ADAS/HSAP
2
2
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: initial HSAP platform

Browse Source 
Huaxu Sentinel Active Safety Platform with embedded algorithm code,
Docker Compose setup, and vendored dataset scaffolds for clone-and-run.

Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
Chengfang Lu
2026-05-25 16:59:59 +08:00
commit 7c43b44c57
1619 changed files with 373355 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

166
algorithms/dms_yolo/code/docs/en/datasets/detect/african-wildlife.md Normal file
View File

@@ -0,0 +1,166 @@
---
comments: true
description: Explore our African Wildlife Dataset featuring images of buffalo, elephant, rhino, and zebra for training computer vision models. Ideal for research and conservation.
keywords: African Wildlife Dataset, South African animals, object detection, computer vision, YOLO26, wildlife research, conservation, dataset
---
# African Wildlife Dataset
This dataset showcases four common animal classes typically found in South African nature reserves. It includes images of African wildlife such as buffalo, elephant, rhino, and zebra, providing valuable insights into their characteristics. Essential for training [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv) algorithms, this dataset aids in identifying animals in various habitats, from zoos to forests, and supports wildlife research.
<p align="center">
<br>
<iframe loading="lazy" width="720" height="405" src="https://www.youtube.com/embed/biIW5Z6GYl0"
title="YouTube video player" frameborder="0"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen>
</iframe>
<br>
<strong>Watch:</strong> African Wildlife Animals Detection using Ultralytics YOLO26
</p>
## Dataset Structure
The African wildlife objects detection dataset is split into three subsets:
- **Training set**: Contains 1052 images, each with corresponding annotations.
- **Validation set**: Includes 225 images, each with paired annotations.
- **Testing set**: Comprises 227 images, each with paired annotations.
## Applications
This dataset can be applied in various computer vision tasks such as [object detection](https://www.ultralytics.com/glossary/object-detection), object tracking, and research. Specifically, it can be used to train and evaluate models for identifying African wildlife objects in images, which can have applications in wildlife conservation, ecological research, and monitoring efforts in natural reserves and protected areas. Additionally, it can serve as a valuable resource for educational purposes, enabling students and researchers to study and understand the characteristics and behaviors of different animal species.
## Dataset YAML
A YAML (Yet Another Markup Language) file defines the dataset configuration, including paths, classes, and other pertinent details. For the African wildlife dataset, the `african-wildlife.yaml` file is located at [https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/african-wildlife.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/african-wildlife.yaml).
!!! example "ultralytics/cfg/datasets/african-wildlife.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/african-wildlife.yaml"
```
## Usage
To train a YOLO26n model on the African wildlife dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, use the provided code samples. For a comprehensive list of available parameters, refer to the model's [Training](../../modes/train.md) page.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="african-wildlife.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=african-wildlife.yaml model=yolo26n.pt epochs=100 imgsz=640
```
!!! example "Inference Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("path/to/best.pt") # load an African wildlife fine-tuned model
# Inference using the model
results = model.predict("https://ultralytics.com/assets/african-wildlife-sample.jpg")
```
=== "CLI"
```bash
# Start prediction with a finetuned *.pt model
yolo detect predict model='path/to/best.pt' imgsz=640 source="https://ultralytics.com/assets/african-wildlife-sample.jpg"
```
## Sample Images and Annotations
The African wildlife dataset comprises a wide variety of images showcasing diverse animal species and their natural habitats. Below are examples of images from the dataset, each accompanied by its corresponding annotations.
![African wildlife dataset sample image](https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/african-wildlife-dataset-sample.avif)
- **Mosaiced Image**: Here, we present a training batch consisting of mosaiced dataset images. Mosaicing, a training technique, combines multiple images into one, enriching batch diversity. This method helps enhance the model's ability to generalize across different object sizes, aspect ratios, and contexts.
This example illustrates the variety and complexity of images in the African wildlife dataset, emphasizing the benefits of including mosaicing during the training process.
## Citations, License and Acknowledgments
We'd like to thank the original dataset author, [Bianca Ferreira](https://www.kaggle.com/biancaferreira/datasets), for releasing this dataset to the community. The Ultralytics team has updated and adapted it internally so it can be used seamlessly with [Ultralytics YOLO](https://www.ultralytics.com/yolo) models. This dataset is available under the [AGPL-3.0 License](https://github.com/ultralytics/ultralytics/blob/main/LICENSE).
If you use this dataset in your research, please cite it using the mentioned details:
!!! quote ""
=== "BibTeX"
```bibtex
@dataset{Ferreira_African_Wildlife_Ultralytics_Adaptation_2024,
author = {Ferreira, Bianca},
title = {African Wildlife Detection Dataset (Ultralytics YOLO Adaptation)},
url = {https://docs.ultralytics.com/datasets/detect/african-wildlife/},
note = {Original dataset by Bianca Ferreira; adapted for Ultralytics YOLO by Glenn Jocher and Muhammad Rizwan Munawar},
license = {AGPL-3.0},
version = {1.0.0},
year = {2024}
}
```
## FAQ
### What is the African Wildlife Dataset, and how can it be used in computer vision projects?
The African Wildlife Dataset includes images of four common animal species found in South African nature reserves: buffalo, elephant, rhino, and zebra. It is a valuable resource for training computer vision algorithms in object detection and animal identification. The dataset supports various tasks like object tracking, research, and conservation efforts. For more information on its structure and applications, refer to the [Dataset Structure](#dataset-structure) section and [Applications](#applications) of the dataset.
### How do I train a YOLO26 model using the African Wildlife Dataset?
You can train a YOLO26 model on the African Wildlife Dataset by using the `african-wildlife.yaml` configuration file. Below is an example of how to train the YOLO26n model for 100 epochs with an image size of 640:
!!! example
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="african-wildlife.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=african-wildlife.yaml model=yolo26n.pt epochs=100 imgsz=640
```
For additional training parameters and options, refer to the [Training](../../modes/train.md) documentation.
### Where can I find the YAML configuration file for the African Wildlife Dataset?
The YAML configuration file for the African Wildlife Dataset, named `african-wildlife.yaml`, can be found at [this GitHub link](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/african-wildlife.yaml). This file defines the dataset configuration, including paths, classes, and other details crucial for training [machine learning](https://www.ultralytics.com/glossary/machine-learning-ml) models. See the [Dataset YAML](#dataset-yaml) section for more details.
### Can I see sample images and annotations from the African Wildlife Dataset?
Yes, the African Wildlife Dataset includes a wide variety of images showcasing diverse animal species in their natural habitats. You can view sample images and their corresponding annotations in the [Sample Images and Annotations](#sample-images-and-annotations) section. This section also illustrates the use of mosaicing technique to combine multiple images into one for enriched batch diversity, enhancing the model's generalization ability.
### How can the African Wildlife Dataset be used to support wildlife conservation and research?
The African Wildlife Dataset is ideal for supporting wildlife conservation and research by enabling the training and evaluation of models to identify African wildlife in different habitats. These models can assist in [monitoring animal populations](https://docs.ultralytics.com/solutions/), studying their behavior, and recognizing conservation needs. Additionally, the dataset can be utilized for educational purposes, helping students and researchers understand the characteristics and behaviors of different animal species. More details can be found in the [Applications](#applications) section.

153
algorithms/dms_yolo/code/docs/en/datasets/detect/argoverse.md Normal file
View File

@@ -0,0 +1,153 @@
---
comments: true
description: Explore the comprehensive Argoverse dataset by Argo AI for 3D tracking, motion forecasting, and stereo depth estimation in autonomous driving research.
keywords: Argoverse dataset, autonomous driving, 3D tracking, motion forecasting, stereo depth estimation, Argo AI, LiDAR point clouds, high-resolution images, HD maps
---
# Argoverse Dataset
The [Argoverse](https://www.argoverse.org/) dataset is a collection of data designed to support research in autonomous driving tasks, such as 3D tracking, motion forecasting, and stereo depth estimation. Developed by Argo AI, the dataset provides a wide range of high-quality sensor data, including high-resolution images, LiDAR point clouds, and map data.
!!! note
The Argoverse dataset `*.zip` file required for training was removed from Amazon S3 after the shutdown of Argo AI by Ford, but we have made it available for manual download on [Google Drive](https://drive.google.com/file/d/1st9qW3BeIwQsnR0t8mRpvbsSWIo16ACi/view?usp=drive_link).
## Key Features
- Argoverse contains over 290K labeled 3D object tracks and 5 million object instances across 1,263 distinct scenes.
- The dataset includes high-resolution camera images, LiDAR point clouds, and richly annotated HD maps.
- Annotations include 3D bounding boxes for objects, object tracks, and trajectory information.
- Argoverse provides multiple subsets for different tasks, such as 3D tracking, motion forecasting, and stereo depth estimation.
## Dataset Structure
The Argoverse dataset is organized into three main subsets:
1. **Argoverse 3D Tracking**: This subset contains 113 scenes with over 290K labeled 3D object tracks, focusing on 3D object tracking tasks. It includes LiDAR point clouds, camera images, and sensor calibration information.
2. **Argoverse Motion Forecasting**: This subset consists of 324K vehicle trajectories collected from 60 hours of driving data, suitable for motion forecasting tasks.
3. **Argoverse Stereo Depth Estimation**: This subset is designed for stereo depth estimation tasks and includes over 10K stereo image pairs with corresponding LiDAR point clouds for ground truth depth estimation.
## Applications
The Argoverse dataset is widely used for training and evaluating [deep learning](https://www.ultralytics.com/glossary/deep-learning-dl) models in autonomous driving tasks such as 3D object tracking, motion forecasting, and stereo depth estimation. The dataset's diverse set of sensor data, object annotations, and map information make it a valuable resource for researchers and practitioners in the field of autonomous driving.
## Dataset YAML
A YAML (Yet Another Markup Language) file is used to define the dataset configuration. It contains information about the dataset's paths, classes, and other relevant information. For the case of the Argoverse dataset, the `Argoverse.yaml` file is maintained at [https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/Argoverse.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/Argoverse.yaml).
!!! example "ultralytics/cfg/datasets/Argoverse.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/Argoverse.yaml"
```
## Usage
To train a YOLO26n model on the Argoverse dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, you can use the following code snippets. For a comprehensive list of available arguments, refer to the model [Training](../../modes/train.md) page.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="Argoverse.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=Argoverse.yaml model=yolo26n.pt epochs=100 imgsz=640
```
## Sample Data and Annotations
The Argoverse dataset contains a diverse set of sensor data, including camera images, LiDAR point clouds, and HD map information, providing rich context for autonomous driving tasks. Here are some examples of data from the dataset, along with their corresponding annotations:
![Argoverse dataset 3D tracking sample with vehicle annotations](https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/argoverse-3d-tracking-sample.avif)
- **Argoverse 3D Tracking**: This image demonstrates an example of 3D object tracking, where objects are annotated with 3D bounding boxes. The dataset provides LiDAR point clouds and camera images to facilitate the development of models for this task.
The example showcases the variety and complexity of the data in the Argoverse dataset and highlights the importance of high-quality sensor data for autonomous driving tasks.
## Citations and Acknowledgments
If you use the Argoverse dataset in your research or development work, please cite the following paper:
!!! quote ""
=== "BibTeX"
```bibtex
@inproceedings{chang2019argoverse,
title={Argoverse: 3D Tracking and Forecasting with Rich Maps},
author={Chang, Ming-Fang and Lambert, John and Sangkloy, Patsorn and Singh, Jagjeet and Bak, Slawomir and Hartnett, Andrew and Wang, Dequan and Carr, Peter and Lucey, Simon and Ramanan, Deva and others},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={8748--8757},
year={2019}
}
```
We would like to acknowledge Argo AI for creating and maintaining the Argoverse dataset as a valuable resource for the autonomous driving research community. For more information about the Argoverse dataset and its creators, visit the [Argoverse dataset website](https://www.argoverse.org/).
## FAQ
### What is the Argoverse dataset and its key features?
The [Argoverse](https://www.argoverse.org/) dataset, developed by Argo AI, supports autonomous driving research. It includes over 290K labeled 3D object tracks and 5 million object instances across 1,263 distinct scenes. The dataset provides high-resolution camera images, LiDAR point clouds, and annotated HD maps, making it valuable for tasks like 3D tracking, motion forecasting, and stereo depth estimation.
### How can I train an Ultralytics YOLO model using the Argoverse dataset?
To train a YOLO26 model with the Argoverse dataset, use the provided YAML configuration file and the following code:
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="Argoverse.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=Argoverse.yaml model=yolo26n.pt epochs=100 imgsz=640
```
For a detailed explanation of the arguments, refer to the model [Training](../../modes/train.md) page.
### What types of data and annotations are available in the Argoverse dataset?
The Argoverse dataset includes various sensor data types such as high-resolution camera images, LiDAR point clouds, and HD map data. Annotations include 3D bounding boxes, object tracks, and trajectory information. These comprehensive annotations are essential for accurate model training in tasks like 3D object tracking, motion forecasting, and stereo depth estimation.
### How is the Argoverse dataset structured?
The dataset is divided into three main subsets:
1. **Argoverse 3D Tracking**: Contains 113 scenes with over 290K labeled 3D object tracks, focusing on 3D object tracking tasks. It includes LiDAR point clouds, camera images, and sensor calibration information.
2. **Argoverse Motion Forecasting**: Consists of 324K vehicle trajectories collected from 60 hours of driving data, suitable for motion forecasting tasks.
3. **Argoverse Stereo Depth Estimation**: Includes over 10K stereo image pairs with corresponding LiDAR point clouds for ground truth depth estimation.
### Where can I download the Argoverse dataset now that it has been removed from Amazon S3?
The Argoverse dataset `*.zip` file, previously available on Amazon S3, can now be manually downloaded from [Google Drive](https://drive.google.com/file/d/1st9qW3BeIwQsnR0t8mRpvbsSWIo16ACi/view?usp=drive_link).
### What is the YAML configuration file used for with the Argoverse dataset?
A YAML file contains the dataset's paths, classes, and other essential information. For the Argoverse dataset, the configuration file, `Argoverse.yaml`, can be found at the following link: [Argoverse.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/Argoverse.yaml).
For more information about YAML configurations, see our [datasets](../index.md) guide.

198
algorithms/dms_yolo/code/docs/en/datasets/detect/brain-tumor.md Normal file
View File

@@ -0,0 +1,198 @@
---
comments: true
description: Explore the brain tumor detection dataset with MRI/CT images. Essential for training AI models for early diagnosis and treatment planning.
keywords: brain tumor dataset, MRI scans, CT scans, brain tumor detection, medical imaging, AI in healthcare, computer vision, early diagnosis, treatment planning
---
# Brain Tumor Dataset
<a href="https://colab.research.google.com/github/ultralytics/notebooks/blob/main/notebooks/how-to-train-ultralytics-yolo-on-brain-tumor-detection-dataset.ipynb"><img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open Brain Tumor Dataset In Colab"></a>
A brain tumor detection dataset consists of medical images from MRI or CT scans, containing information about brain tumor presence, location, and characteristics. This dataset is essential for training [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv) algorithms to automate brain tumor identification, aiding in early diagnosis and treatment planning in [healthcare applications](https://www.ultralytics.com/solutions/ai-in-healthcare).
<p align="center">
<br>
<iframe loading="lazy" width="720" height="405" src="https://www.youtube.com/embed/ogTBBD8McRk"
title="YouTube video player" frameborder="0"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen>
</iframe>
<br>
<strong>Watch:</strong> Brain Tumor Detection using Ultralytics Platform
</p>
## Dataset Structure
The brain tumor dataset is divided into two subsets:
- **Training set**: Consisting of 893 images, each accompanied by corresponding annotations.
- **Testing set**: Comprising 223 images, with annotations paired for each one.
The dataset contains two classes:
- **Negative**: Images without brain tumors
- **Positive**: Images with brain tumors
## Applications
The application of brain tumor detection using computer vision enables [early diagnosis](https://www.ultralytics.com/blog/ai-and-radiology-a-new-era-of-precision-and-efficiency), treatment planning, and monitoring of tumor progression. By analyzing medical imaging data like MRI or CT scans, [computer vision systems](https://docs.ultralytics.com/tasks/detect/) assist in accurately identifying brain tumors, aiding in timely medical intervention and personalized treatment strategies.
Medical professionals can leverage this technology to:
- Reduce diagnostic time and improve accuracy
- Assist in surgical planning by precisely locating tumors
- Monitor treatment effectiveness over time
- Support research in oncology and neurology
## Dataset YAML
A YAML (Yet Another Markup Language) file is used to define the dataset configuration. It contains information about the dataset's paths, classes, and other relevant information. In the case of the brain tumor dataset, the `brain-tumor.yaml` file is maintained at [https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/brain-tumor.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/brain-tumor.yaml).
!!! example "ultralytics/cfg/datasets/brain-tumor.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/brain-tumor.yaml"
```
## Usage
To train a [YOLO26](https://docs.ultralytics.com/models/yolo26/) model on the brain tumor dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, utilize the provided code snippets. For a detailed list of available arguments, consult the model's [Training](../../modes/train.md) page.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="brain-tumor.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=brain-tumor.yaml model=yolo26n.pt epochs=100 imgsz=640
```
!!! example "Inference Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("path/to/best.pt") # load a brain-tumor fine-tuned model
# Inference using the model
results = model.predict("https://ultralytics.com/assets/brain-tumor-sample.jpg")
```
=== "CLI"
```bash
# Start prediction with a finetuned *.pt model
yolo detect predict model='path/to/best.pt' imgsz=640 source="https://ultralytics.com/assets/brain-tumor-sample.jpg"
```
## Sample Images and Annotations
The brain tumor dataset encompasses a wide array of medical images featuring brain scans with and without tumors. Presented below are examples of images from the dataset, accompanied by their respective annotations.
![Brain tumor dataset sample image](https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/brain-tumor-dataset-sample-image.avif)
- **Mosaiced Image**: Displayed here is a training batch comprising mosaiced dataset images. Mosaicing, a training technique, consolidates multiple images into one, enhancing batch diversity. This approach aids in improving the model's capacity to generalize across various tumor sizes, shapes, and locations within brain scans.
This example highlights the diversity and intricacy of images within the brain tumor dataset, underscoring the advantages of incorporating mosaicing during the training phase for [medical image analysis](https://www.ultralytics.com/blog/using-yolo11-for-tumor-detection-in-medical-imaging).
## Citations and Acknowledgments
The dataset has been made available under the [AGPL-3.0 License](https://github.com/ultralytics/ultralytics/blob/main/LICENSE).
If you use this dataset in your research or development work, please cite it appropriately:
!!! quote ""
=== "BibTeX"
```bibtex
@dataset{Ultralytics_Brain_Tumor_Dataset_2023,
author = {Ultralytics},
title = {Brain Tumor Detection Dataset},
year = {2023},
publisher = {Ultralytics},
url = {https://docs.ultralytics.com/datasets/detect/brain-tumor/}
}
```
## FAQ
### What is the structure of the brain tumor dataset available in Ultralytics documentation?
The brain tumor dataset is divided into two subsets: the **training set** consists of 893 images with corresponding annotations, while the **testing set** comprises 223 images with paired annotations. This structured division aids in developing robust and accurate computer vision models for detecting brain tumors. For more information on the dataset structure, visit the [Dataset Structure](#dataset-structure) section.
### How can I train a YOLO26 model on the brain tumor dataset using Ultralytics?
You can train a YOLO26 model on the brain tumor dataset for 100 epochs with an image size of 640px using both Python and CLI methods. Below are the examples for both:
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="brain-tumor.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=brain-tumor.yaml model=yolo26n.pt epochs=100 imgsz=640
```
For a detailed list of available arguments, refer to the [Training](../../modes/train.md) page.
### What are the benefits of using the brain tumor dataset for AI in healthcare?
Using the brain tumor dataset in AI projects enables early diagnosis and treatment planning for brain tumors. It helps in automating brain tumor identification through computer vision, facilitating accurate and timely medical interventions, and supporting personalized treatment strategies. This application holds significant potential in improving patient outcomes and medical efficiencies. For more insights on AI applications in healthcare, see [Ultralytics' healthcare solutions](https://www.ultralytics.com/solutions/ai-in-healthcare).
### How do I perform inference using a fine-tuned YOLO26 model on the brain tumor dataset?
Inference using a fine-tuned YOLO26 model can be performed with either Python or CLI approaches. Here are the examples:
!!! example "Inference Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("path/to/best.pt") # load a brain-tumor fine-tuned model
# Inference using the model
results = model.predict("https://ultralytics.com/assets/brain-tumor-sample.jpg")
```
=== "CLI"
```bash
# Start prediction with a finetuned *.pt model
yolo detect predict model='path/to/best.pt' imgsz=640 source="https://ultralytics.com/assets/brain-tumor-sample.jpg"
```
### Where can I find the YAML configuration for the brain tumor dataset?
The YAML configuration file for the brain tumor dataset can be found at [brain-tumor.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/brain-tumor.yaml). This file includes paths, classes, and additional relevant information necessary for training and evaluating models on this dataset.

173
algorithms/dms_yolo/code/docs/en/datasets/detect/coco.md Normal file
View File

@@ -0,0 +1,173 @@
---
comments: true
description: Explore the COCO dataset for object detection and segmentation. Learn about its structure, usage, pretrained models, and key features.
keywords: COCO dataset, object detection, segmentation, benchmarking, computer vision, pose estimation, YOLO models, COCO annotations
---
# COCO Dataset
The [COCO](https://cocodataset.org/#home) (Common Objects in Context) dataset is a large-scale object detection, segmentation, and captioning dataset. It is designed to encourage research on a wide variety of object categories and is commonly used for benchmarking [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv) models. It is an essential dataset for researchers and developers working on object detection, segmentation, and pose estimation tasks.
<p align="center">
<br>
<iframe loading="lazy" width="720" height="405" src="https://www.youtube.com/embed/uDrn9QZJ2lk"
title="YouTube video player" frameborder="0"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen>
</iframe>
<br>
<strong>Watch:</strong> Ultralytics COCO Dataset Overview
</p>
## COCO Pretrained Models
{% include "macros/yolo-det-perf.md" %}
## Key Features
- COCO contains 330K images, with 200K images having annotations for object detection, segmentation, and captioning tasks.
- The dataset comprises 80 object categories, including common objects like cars, bicycles, and animals, as well as more specific categories such as umbrellas, handbags, and sports equipment.
- Annotations include object bounding boxes, segmentation masks, and captions for each image.
- COCO provides standardized evaluation metrics like [mean Average Precision](https://www.ultralytics.com/glossary/mean-average-precision-map) (mAP) for object detection, and mean Average [Recall](https://www.ultralytics.com/glossary/recall) (mAR) for segmentation tasks, making it suitable for comparing model performance.
## Dataset Structure
The COCO dataset is split into three subsets:
1. **Train2017**: This subset contains 118K images for training object detection, segmentation, and captioning models.
2. **Val2017**: This subset has 5K images used for validation purposes during model training.
3. **Test2017**: This subset consists of 20K images used for testing and benchmarking the trained models. Ground truth annotations for this subset are not publicly available, and the results are submitted to the [COCO evaluation server](https://codalab.lisn.upsaclay.fr/competitions/7384) for performance evaluation.
## Applications
The COCO dataset is widely used for training and evaluating [deep learning](https://www.ultralytics.com/glossary/deep-learning-dl) models in object detection (such as [Ultralytics YOLO](../../models/yolo26.md), [Faster R-CNN](https://arxiv.org/abs/1506.01497), and [SSD](https://arxiv.org/abs/1512.02325)), [instance segmentation](https://www.ultralytics.com/glossary/instance-segmentation) (such as [Mask R-CNN](https://arxiv.org/abs/1703.06870)), and keypoint detection (such as [OpenPose](https://arxiv.org/abs/1812.08008)). The dataset's diverse set of object categories, large number of annotated images, and standardized evaluation metrics make it an essential resource for computer vision researchers and practitioners.
## Dataset YAML
A YAML (Yet Another Markup Language) file is used to define the dataset configuration. It contains information about the dataset's paths, classes, and other relevant information. In the case of the COCO dataset, the `coco.yaml` file is maintained at [https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/coco.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/coco.yaml).
!!! example "ultralytics/cfg/datasets/coco.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/coco.yaml"
```
## Usage
To train a YOLO26n model on the COCO dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, you can use the following code snippets. For a comprehensive list of available arguments, refer to the model [Training](../../modes/train.md) page.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="coco.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=coco.yaml model=yolo26n.pt epochs=100 imgsz=640
```
## Sample Images and Annotations
The COCO dataset contains a diverse set of images with various object categories and complex scenes. Here are some examples of images from the dataset, along with their corresponding annotations:
![COCO dataset mosaic training batch with object detection](https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/mosaiced-coco-dataset-sample.avif)
- **Mosaiced Image**: This image demonstrates a training batch composed of mosaiced dataset images. Mosaicing is a technique used during training that combines multiple images into a single image to increase the variety of objects and scenes within each training batch. This helps improve the model's ability to generalize to different object sizes, aspect ratios, and contexts.
The example showcases the variety and complexity of the images in the COCO dataset and the benefits of using mosaicing during the training process.
## Citations and Acknowledgments
If you use the COCO dataset in your research or development work, please cite the following paper:
!!! quote ""
=== "BibTeX"
```bibtex
@misc{lin2015microsoft,
title={Microsoft COCO: Common Objects in Context},
author={Tsung-Yi Lin and Michael Maire and Serge Belongie and Lubomir Bourdev and Ross Girshick and James Hays and Pietro Perona and Deva Ramanan and C. Lawrence Zitnick and Piotr Dollár},
year={2015},
eprint={1405.0312},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
```
We would like to acknowledge the COCO Consortium for creating and maintaining this valuable resource for the computer vision community. For more information about the COCO dataset and its creators, visit the [COCO dataset website](https://cocodataset.org/#home).
## FAQ
### What is the COCO dataset and why is it important for computer vision?
The [COCO dataset](https://cocodataset.org/#home) (Common Objects in Context) is a large-scale dataset used for [object detection](https://www.ultralytics.com/glossary/object-detection), segmentation, and captioning. It contains 330K images with detailed annotations for 80 object categories, making it essential for benchmarking and training computer vision models. Researchers use COCO due to its diverse categories and standardized evaluation metrics like mean Average [Precision](https://www.ultralytics.com/glossary/precision) (mAP).
### How can I train a YOLO model using the COCO dataset?
To train a YOLO26 model using the COCO dataset, you can use the following code snippets:
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="coco.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=coco.yaml model=yolo26n.pt epochs=100 imgsz=640
```
Refer to the [Training page](../../modes/train.md) for more details on available arguments.
### What are the key features of the COCO dataset?
The COCO dataset includes:
- 330K images, with 200K annotated for object detection, segmentation, and captioning.
- 80 object categories ranging from common items like cars and animals to specific ones like handbags and sports equipment.
- Standardized evaluation metrics for object detection (mAP) and segmentation (mean Average Recall, mAR).
- **Mosaicing** technique in training batches to enhance model generalization across various object sizes and contexts.
### Where can I find pretrained YOLO26 models trained on the COCO dataset?
Pretrained YOLO26 models on the COCO dataset can be downloaded from the links provided in the documentation. Examples include:
- [YOLO26n](https://github.com/ultralytics/assets/releases/download/v8.4.0/yolo26n.pt)
- [YOLO26s](https://github.com/ultralytics/assets/releases/download/v8.4.0/yolo26s.pt)
- [YOLO26m](https://github.com/ultralytics/assets/releases/download/v8.4.0/yolo26m.pt)
- [YOLO26l](https://github.com/ultralytics/assets/releases/download/v8.4.0/yolo26l.pt)
- [YOLO26x](https://github.com/ultralytics/assets/releases/download/v8.4.0/yolo26x.pt)
These models vary in size, mAP, and inference speed, providing options for different performance and resource requirements.
### How is the COCO dataset structured and how do I use it?
The COCO dataset is split into three subsets:
1. **Train2017**: 118K images for training.
2. **Val2017**: 5K images for validation during training.
3. **Test2017**: 20K images for benchmarking trained models. Results need to be submitted to the [COCO evaluation server](https://codalab.lisn.upsaclay.fr/competitions/7384) for performance evaluation.
The dataset's YAML configuration file is available at [coco.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/coco.yaml), which defines paths, classes, and dataset details.

234
algorithms/dms_yolo/code/docs/en/datasets/detect/coco12-formats.md Normal file
View File

@@ -0,0 +1,234 @@
---
comments: true
description: Explore the Ultralytics COCO12-Formats dataset, a test dataset featuring all 12 supported image formats (AVIF, BMP, DNG, HEIC, JP2, JPEG, JPG, MPO, PNG, TIF, TIFF, WebP) for validating image loading pipelines.
keywords: COCO12-Formats, Ultralytics, dataset, image formats, object detection, YOLO, AVIF, BMP, DNG, HEIC, JP2, JPEG, PNG, TIFF, WebP, MPO
---
# COCO12-Formats Dataset
## Introduction
The [Ultralytics](https://www.ultralytics.com/) COCO12-Formats dataset is a specialized test dataset designed to validate image loading across all 12 supported image format extensions. It contains 12 images (6 for training, 6 for validation), each saved in a different format to ensure comprehensive testing of the image loading pipeline.
This dataset is invaluable for:
- **Testing image format support**: Verify that all supported formats load correctly
- **CI/CD pipelines**: Automated testing of format compatibility
- **Debugging**: Isolate format-specific issues in training pipelines
- **Development**: Validate new format additions or changes
## Supported Formats
The dataset includes one image for each of the 12 supported format extensions defined in `ultralytics/data/utils.py`:
| Format | Extension | Description | Train/Val |
| ------ | --------- | ------------------------------------ | --------- |
| AVIF | `.avif` | AV1 Image File Format (modern) | Train |
| BMP | `.bmp` | Bitmap - uncompressed raster format | Train |
| DNG | `.dng` | Digital Negative - Adobe RAW format | Train |
| HEIC | `.heic` | High Efficiency Image Coding | Train |
| JPEG | `.jpeg` | JPEG with full extension | Train |
| JPG | `.jpg` | JPEG with short extension | Train |
| JP2 | `.jp2` | JPEG 2000 - medical/geospatial | Val |
| MPO | `.mpo` | Multi-Picture Object (stereo images) | Val |
| PNG | `.png` | Portable Network Graphics | Val |
| TIF | `.tif` | TIFF with short extension | Val |
| TIFF | `.tiff` | Tagged Image File Format | Val |
| WebP | `.webp` | Modern web image format | Val |
## Dataset Structure
```
coco12-formats/
├── images/
│ ├── train/ # 6 images (avif, bmp, dng, heic, jpeg, jpg)
│ └── val/ # 6 images (jp2, mpo, png, tif, tiff, webp)
├── labels/
│ ├── train/ # Corresponding YOLO format labels
│ └── val/
└── coco12-formats.yaml # Dataset configuration
```
## Dataset YAML
The COCO12-Formats dataset is configured using a YAML file that defines dataset paths and class names. You can review the official `coco12-formats.yaml` file in the [Ultralytics GitHub repository](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/coco12-formats.yaml).
!!! example "ultralytics/cfg/datasets/coco12-formats.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/coco12-formats.yaml"
```
## Dataset Generation
The dataset can be generated using the provided script that converts source images from COCO8 and COCO128 to all supported formats:
```python
from ultralytics.data.scripts.generate_coco12_formats import generate_coco12_formats
# Generate the dataset
generate_coco12_formats()
```
### Requirements
Some formats require additional dependencies:
```bash
pip install pillow pillow-heif pillow-avif-plugin
```
#### AVIF System Library (Optional)
For OpenCV to read AVIF files directly, `libavif` must be installed **before** building OpenCV:
=== "macOS"
```bash
brew install libavif
```
=== "Ubuntu/Debian"
```bash
sudo apt install libavif-dev libavif-bin
```
=== "From Source"
```bash
git clone -b v1.2.1 https://github.com/AOMediaCodec/libavif.git
cd libavif
cmake -B build -DAVIF_CODEC_AOM=SYSTEM -DAVIF_BUILD_APPS=ON
cmake --build build --config Release --parallel
sudo cmake --install build
```
!!! note
The pip-installed `opencv-python` package may not include AVIF support since it's pre-built. Ultralytics uses Pillow with `pillow-avif-plugin` as a fallback for AVIF images when OpenCV lacks support.
## Usage
To train a YOLO model on the COCO12-Formats dataset, use the following examples:
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a pretrained YOLO model
model = YOLO("yolo26n.pt")
# Train on COCO12-Formats to test all image formats
results = model.train(data="coco12-formats.yaml", epochs=1, imgsz=640)
```
=== "CLI"
```bash
# Train YOLO on COCO12-Formats
yolo detect train data=coco12-formats.yaml model=yolo26n.pt epochs=1 imgsz=640
```
## Format-Specific Notes
### AVIF (AV1 Image File Format)
AVIF is a modern image format based on the AV1 video codec, offering excellent compression. Requires `pillow-avif-plugin`:
```bash
pip install pillow-avif-plugin
```
### DNG (Digital Negative)
DNG is Adobe's open RAW format based on TIFF. For testing purposes, the dataset uses TIFF-based files with the `.dng` extension.
### JP2 (JPEG 2000)
JPEG 2000 is a wavelet-based image compression standard offering better compression and quality than traditional JPEG. Commonly used in medical imaging (DICOM), geospatial applications, and digital cinema. Natively supported by both OpenCV and Pillow.
### MPO (Multi-Picture Object)
MPO files are used for stereoscopic (3D) images. The dataset stores standard JPEG data with the `.mpo` extension for format testing.
### HEIC (High Efficiency Image Coding)
HEIC requires the `pillow-heif` package for proper encoding:
```bash
pip install pillow-heif
```
## Use Cases
### CI/CD Testing
```python
from ultralytics import YOLO
def test_all_image_formats():
"""Test that all image formats load correctly."""
model = YOLO("yolo26n.pt")
results = model.train(data="coco12-formats.yaml", epochs=1, imgsz=64)
assert results is not None
```
### Format Validation
```python
from pathlib import Path
from ultralytics.data.utils import IMG_FORMATS
# Verify all formats are represented
dataset_dir = Path("datasets/coco12-formats/images")
found_formats = {f.suffix[1:].lower() for f in dataset_dir.rglob("*.*")}
assert found_formats == IMG_FORMATS, f"Missing formats: {IMG_FORMATS - found_formats}"
```
## Citations and Acknowledgments
If you use the COCO dataset in your research, please cite:
!!! quote ""
=== "BibTeX"
```bibtex
@misc{lin2015microsoft,
title={Microsoft COCO: Common Objects in Context},
author={Tsung-Yi Lin and Michael Maire and Serge Belongie and Lubomir Bourdev and Ross Girshick and James Hays and Pietro Perona and Deva Ramanan and C. Lawrence Zitnick and Piotr Doll{\'a}r},
year={2015},
eprint={1405.0312},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
```
## FAQ
### What Is the COCO12-Formats Dataset Used For?
The COCO12-Formats dataset is designed for testing image format compatibility in Ultralytics YOLO training pipelines. It ensures all 12 supported image formats (AVIF, BMP, DNG, HEIC, JP2, JPEG, JPG, MPO, PNG, TIF, TIFF, WebP) load and process correctly.
### Why Test Multiple Image Formats?
Different image formats have unique characteristics (compression, bit depth, color spaces). Testing all formats ensures:
- Robust image loading code
- Compatibility across diverse datasets
- Early detection of format-specific bugs
### Which Formats Require Special Dependencies?
- **AVIF**: Requires `pillow-avif-plugin`
- **HEIC**: Requires `pillow-heif`
### Can I Add New Format Tests?
Yes! Modify the `generate_coco12_formats.py` script to include additional formats. Ensure you also update `IMG_FORMATS` in `ultralytics/data/utils.py`.

154
algorithms/dms_yolo/code/docs/en/datasets/detect/coco128.md Normal file
View File

@@ -0,0 +1,154 @@
---
comments: true
description: Explore the Ultralytics COCO128 dataset, a versatile and manageable set of 128 images perfect for testing object detection models and training pipelines.
keywords: COCO128, Ultralytics, dataset, object detection, YOLO26, training, validation, machine learning, computer vision
---
# COCO128 Dataset
## Introduction
[Ultralytics](https://www.ultralytics.com/) COCO128 is a small, but versatile [object detection](https://www.ultralytics.com/glossary/object-detection) dataset composed of the first 128 images of the COCO train 2017 set. This dataset is ideal for testing and debugging object detection models, or for experimenting with new detection approaches. With 128 images, it is small enough to be easily manageable, yet diverse enough to test training pipelines for errors and act as a sanity check before training larger datasets.
<p align="center">
<br>
<iframe loading="lazy" width="720" height="405" src="https://www.youtube.com/embed/uDrn9QZJ2lk"
title="YouTube video player" frameborder="0"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen>
</iframe>
<br>
<strong>Watch:</strong> Ultralytics COCO Dataset Overview
</p>
This dataset is intended for use with [Ultralytics Platform](https://platform.ultralytics.com/) and [YOLO26](https://github.com/ultralytics/ultralytics).
## Dataset YAML
A YAML (Yet Another Markup Language) file is used to define the dataset configuration. It contains information about the dataset's paths, classes, and other relevant information. In the case of the COCO128 dataset, the `coco128.yaml` file is maintained at [https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/coco128.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/coco128.yaml).
!!! example "ultralytics/cfg/datasets/coco128.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/coco128.yaml"
```
## Usage
To train a YOLO26n model on the COCO128 dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, you can use the following code snippets. For a comprehensive list of available arguments, refer to the model [Training](../../modes/train.md) page.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="coco128.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=coco128.yaml model=yolo26n.pt epochs=100 imgsz=640
```
## Sample Images and Annotations
Here are some examples of images from the COCO128 dataset, along with their corresponding annotations:
<img src="https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/mosaiced-training-batch-1.avif" alt="COCO128 object detection dataset mosaic training batch" width="800">
- **Mosaiced Image**: This image demonstrates a training batch composed of mosaiced dataset images. Mosaicing is a technique used during training that combines multiple images into a single image to increase the variety of objects and scenes within each training batch. This helps improve the model's ability to generalize to different object sizes, aspect ratios, and contexts.
The example showcases the variety and complexity of the images in the COCO128 dataset and the benefits of using mosaicing during the training process.
## Citations and Acknowledgments
If you use the COCO dataset in your research or development work, please cite the following paper:
!!! quote ""
=== "BibTeX"
```bibtex
@misc{lin2015microsoft,
title={Microsoft COCO: Common Objects in Context},
author={Tsung-Yi Lin and Michael Maire and Serge Belongie and Lubomir Bourdev and Ross Girshick and James Hays and Pietro Perona and Deva Ramanan and C. Lawrence Zitnick and Piotr Dollár},
year={2015},
eprint={1405.0312},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
```
We would like to acknowledge the COCO Consortium for creating and maintaining this valuable resource for the [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv) community. For more information about the COCO dataset and its creators, visit the [COCO dataset website](https://cocodataset.org/#home).
## FAQ
### What is the Ultralytics COCO128 dataset used for?
The Ultralytics COCO128 dataset is a compact subset containing the first 128 images from the COCO train 2017 dataset. It's primarily used for testing and debugging [object detection](https://www.ultralytics.com/glossary/object-detection) models, experimenting with new detection approaches, and validating training pipelines before scaling to larger datasets. Its manageable size makes it perfect for quick iterations while still providing enough diversity to be a meaningful test case.
### How do I train a YOLO26 model using the COCO128 dataset?
To train a YOLO26 model on the COCO128 dataset, you can use either Python or CLI commands. Here's how:
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a pretrained model
model = YOLO("yolo26n.pt")
# Train the model
results = model.train(data="coco128.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
yolo detect train data=coco128.yaml model=yolo26n.pt epochs=100 imgsz=640
```
For more training options and parameters, refer to the [Training](../../modes/train.md) documentation.
### What are the benefits of using mosaic augmentation with COCO128?
Mosaic augmentation, as shown in the sample images, combines multiple training images into a single composite image. This technique offers several benefits when training with COCO128:
- Increases the variety of objects and contexts within each training batch
- Improves model generalization across different object sizes and aspect ratios
- Enhances detection performance for objects at various scales
- Maximizes the utility of a small dataset by creating more diverse training samples
This technique is particularly valuable for smaller datasets like COCO128, helping models learn more robust features from limited data.
### How does COCO128 compare to other COCO dataset variants?
COCO128 (128 images) sits between [COCO8](../detect/coco8.md) (8 images) and the full [COCO](../detect/coco.md) dataset (118K+ images) in terms of size:
- **COCO8**: Contains just 8 images (4 train, 4 val) - ideal for quick tests and debugging
- **COCO128**: Contains 128 images - balanced between size and diversity
- **Full COCO**: Contains 118K+ training images - comprehensive but resource-intensive
COCO128 provides a good middle ground, offering more diversity than COCO8 while remaining much more manageable than the full COCO dataset for experimentation and initial model development.
### Can I use COCO128 for tasks other than object detection?
While COCO128 is primarily designed for object detection, the dataset's annotations can be adapted for other computer vision tasks:
- **Instance segmentation**: Using the segmentation masks provided in the annotations
- **Keypoint detection**: For images containing people with keypoint annotations
- **Transfer learning**: As a starting point for fine-tuning models for custom tasks
For specialized tasks like [segmentation](../../tasks/segment.md), consider using purpose-built variants like [COCO8-seg](../segment/coco8-seg.md) which include the appropriate annotations.

138
algorithms/dms_yolo/code/docs/en/datasets/detect/coco8-grayscale.md Normal file
View File

@@ -0,0 +1,138 @@
---
comments: true
description: Explore the Ultralytics COCO8-Grayscale dataset, a versatile and manageable set of 8 images perfect for testing object detection models and training pipelines.
keywords: COCO8-Grayscale, Ultralytics, dataset, object detection, YOLO26, training, validation, machine learning, computer vision
---
# COCO8-Grayscale Dataset
## Introduction
The [Ultralytics](https://www.ultralytics.com/) COCO8-Grayscale dataset is a compact yet powerful [object detection](https://www.ultralytics.com/glossary/object-detection) dataset, consisting of the first 8 images from the COCO train 2017 set and converted to grayscale format—4 for training and 4 for validation. This dataset is specifically designed for rapid testing, debugging, and experimentation with [YOLO](https://docs.ultralytics.com/models/yolo26/) grayscale models and training pipelines. Its small size makes it highly manageable, while its diversity ensures it serves as an effective sanity check before scaling up to larger datasets.
<p align="center">
<br>
<iframe loading="lazy" width="720" height="405" src="https://www.youtube.com/embed/yw2Fo6qjJU4"
title="YouTube video player" frameborder="0"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen>
</iframe>
<br>
<strong>Watch:</strong> How to Train Ultralytics YOLO26 on Grayscale Datasets 🚀
</p>
COCO8-Grayscale is fully compatible with [Ultralytics Platform](https://platform.ultralytics.com/) and [YOLO26](../../models/yolo26.md), enabling seamless integration into your computer vision workflows.
## Dataset YAML
The COCO8-Grayscale dataset configuration is defined in a YAML (Yet Another Markup Language) file, which specifies dataset paths, class names, and other essential metadata. You can review the official `coco8-grayscale.yaml` file in the [Ultralytics GitHub repository](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/coco8-grayscale.yaml).
!!! note
To train your RGB images in grayscale, you could simply add `channels: 1` to your dataset YAML file. This converts all images to grayscale during training, enabling you to utilize grayscale benefits without requiring a separate dataset.
!!! example "ultralytics/cfg/datasets/coco8-grayscale.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/coco8-grayscale.yaml"
```
## Usage
To train a YOLO26n model on the COCO8-Grayscale dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, use the following examples. For a full list of training options, see the [YOLO Training documentation](../../modes/train.md).
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a pretrained YOLO26n model
model = YOLO("yolo26n.pt")
# Train the model on COCO8-Grayscale
results = model.train(data="coco8-grayscale.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Train YOLO26n on COCO8-Grayscale using the command line
yolo detect train data=coco8-grayscale.yaml model=yolo26n.pt epochs=100 imgsz=640
```
## Sample Images and Annotations
Below is an example of a mosaiced training batch from the COCO8-Grayscale dataset:
<img src="https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/grayscale-mosaic.avif" alt="COCO8 grayscale dataset mosaic training batch" width="800">
- **Mosaiced Image**: This image illustrates a training batch where multiple dataset images are combined using mosaic augmentation. Mosaic augmentation increases the diversity of objects and scenes within each batch, helping the model generalize better to various object sizes, aspect ratios, and backgrounds.
This technique is especially useful for small datasets like COCO8-Grayscale, as it maximizes the value of each image during training.
## Citations and Acknowledgments
If you use the COCO dataset in your research or development, please cite the following paper:
!!! quote ""
=== "BibTeX"
```bibtex
@misc{lin2015microsoft,
title={Microsoft COCO: Common Objects in Context},
author={Tsung-Yi Lin and Michael Maire and Serge Belongie and Lubomir Bourdev and Ross Girshick and James Hays and Pietro Perona and Deva Ramanan and C. Lawrence Zitnick and Piotr Dollár},
year={2015},
eprint={1405.0312},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
```
Special thanks to the [COCO Consortium](https://cocodataset.org/#home) for their ongoing contributions to the [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv) community.
## FAQ
### What Is the Ultralytics COCO8-Grayscale Dataset Used For?
The Ultralytics COCO8-Grayscale dataset is designed for rapid testing and debugging of [object detection](https://www.ultralytics.com/glossary/object-detection) models. With only 8 images (4 for training, 4 for validation), it is ideal for verifying your [YOLO](https://docs.ultralytics.com/models/yolo26/) training pipelines and ensuring everything works as expected before scaling to larger datasets. Explore the [COCO8-Grayscale YAML configuration](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/coco8-grayscale.yaml) for more details.
### How Do I Train a YOLO26 Model Using the COCO8-Grayscale Dataset?
You can train a YOLO26 model on COCO8-Grayscale using either Python or the CLI:
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a pretrained YOLO26n model
model = YOLO("yolo26n.pt")
# Train the model on COCO8-Grayscale
results = model.train(data="coco8-grayscale.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
yolo detect train data=coco8-grayscale.yaml model=yolo26n.pt epochs=100 imgsz=640
```
For additional training options, refer to the [YOLO Training documentation](../../modes/train.md).
### Why Should I Use Ultralytics Platform for Managing My COCO8-Grayscale Training?
[Ultralytics Platform](https://platform.ultralytics.com/) streamlines dataset management, training, and deployment for [YOLO](https://docs.ultralytics.com/models/yolo26/) models—including COCO8-Grayscale. With features like cloud training, real-time monitoring, and intuitive dataset handling, HUB enables you to launch experiments with a single click and eliminates manual setup hassles. Learn more about [Ultralytics Platform](https://platform.ultralytics.com/) and how it can accelerate your computer vision projects.
### What Are the Benefits of Using Mosaic Augmentation in Training With the COCO8-Grayscale Dataset?
Mosaic augmentation, as used in COCO8-Grayscale training, combines multiple images into one during each batch. This increases the diversity of objects and backgrounds, helping your [YOLO](https://docs.ultralytics.com/models/yolo26/) model generalize better to new scenarios. Mosaic augmentation is especially valuable for small datasets, as it maximizes the information available in each training step. For more on this, see the [training guide](#usage).
### How Can I Validate My YOLO26 Model Trained on the COCO8-Grayscale Dataset?
To validate your YOLO26 model after training on COCO8-Grayscale, use the model's validation commands in either Python or CLI. This evaluates your model's performance using standard metrics. For step-by-step instructions, visit the [YOLO Validation documentation](../../modes/val.md).

146
algorithms/dms_yolo/code/docs/en/datasets/detect/coco8-multispectral.md Normal file
View File

@@ -0,0 +1,146 @@
---
comments: true
description: Explore the Ultralytics COCO8-Multispectral dataset, an enhanced version of COCO8 with interpolated spectral channels, ideal for testing multispectral object detection models and training pipelines.
keywords: COCO8-Multispectral, Ultralytics, dataset, multispectral, object detection, YOLO26, training, validation, machine learning, computer vision
---
# COCO8-Multispectral Dataset
## Introduction
The [Ultralytics](https://www.ultralytics.com/) COCO8-Multispectral dataset is an advanced variant of the original COCO8 dataset, designed to facilitate experimentation with multispectral object detection models. It consists of the same 8 images from the COCO train 2017 set—4 for training and 4 for validation—but with each image transformed into a 10-channel multispectral format. By expanding beyond standard RGB channels, COCO8-Multispectral enables the development and evaluation of models that can leverage richer spectral information.
<p align="center">
<img width="640" src="https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/coco8-multispectral-overview.avif" alt="Multispectral imaging for object detection">
</p>
COCO8-Multispectral is fully compatible with [Ultralytics Platform](https://platform.ultralytics.com/) and [YOLO26](../../models/yolo26.md), ensuring seamless integration into your [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv) workflows.
<p align="center">
<br>
<iframe loading="lazy" width="720" height="405" src="https://www.youtube.com/embed/yw2Fo6qjJU4"
title="YouTube video player" frameborder="0"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen>
</iframe>
<br>
<strong>Watch:</strong> How to Train Ultralytics YOLO26 on Multispectral Datasets | Multi-Channel VisionAI 🚀
</p>
## Dataset Generation
The multispectral images in COCO8-Multispectral were created by interpolating the original RGB images across 10 evenly spaced spectral channels within the visible spectrum. The process includes:
- **Wavelength Assignment**: Assigning nominal wavelengths to the RGB channels—Red: 650 nm, Green: 510 nm, Blue: 475 nm.
- **Interpolation**: Using linear interpolation to estimate pixel values at intermediate wavelengths between 450 nm and 700 nm, resulting in 10 spectral channels.
- **Extrapolation**: Applying extrapolation with SciPy's `interp1d` function to estimate values beyond the original RGB wavelengths, ensuring a complete spectral representation.
This approach simulates a multispectral imaging process, providing a more diverse set of data for model training and evaluation. For further reading on multispectral imaging, see the [Multispectral Imaging Wikipedia article](https://en.wikipedia.org/wiki/Multispectral_imaging).
## Dataset YAML
The COCO8-Multispectral dataset is configured using a YAML file, which defines dataset paths, class names, and essential metadata. You can review the official `coco8-multispectral.yaml` file in the [Ultralytics GitHub repository](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/coco8-multispectral.yaml).
!!! example "ultralytics/cfg/datasets/coco8-multispectral.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/coco8-multispectral.yaml"
```
!!! note
Prepare your TIFF images in `(channel, height, width)` order, saved with `.tiff` or `.tif` extension, and ensure they are `uint8` for use with Ultralytics:
```python
import cv2
import numpy as np
# Create and write 10-channel TIFF
image = np.ones((10, 640, 640), dtype=np.uint8) # CHW-order
cv2.imwritemulti("example.tiff", image)
# Read TIFF
success, frames_list = cv2.imreadmulti("example.tiff")
image = np.stack(frames_list, axis=2)
print(image.shape) # (640, 640, 10) HWC-order for training and inference
```
## Usage
To train a YOLO26n model on the COCO8-Multispectral dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, use the following examples. For a comprehensive list of training options, refer to the [YOLO Training documentation](../../modes/train.md).
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a pretrained YOLO26n model
model = YOLO("yolo26n.pt")
# Train the model on COCO8-Multispectral
results = model.train(data="coco8-multispectral.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Train YOLO26n on COCO8-Multispectral using the command line
yolo detect train data=coco8-multispectral.yaml model=yolo26n.pt epochs=100 imgsz=640
```
For more details on model selection and best practices, explore the [Ultralytics YOLO model documentation](../../models/yolo26.md) and the [YOLO Model Training Tips guide](https://docs.ultralytics.com/guides/model-training-tips/).
## Sample Images and Annotations
Below is an example of a mosaiced training batch from the COCO8-Multispectral dataset:
<img src="https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/coco8-multispectral-mosaic-batch.avif" alt="COCO8 multispectral dataset mosaic training batch" width="800">
- **Mosaiced Image**: This image demonstrates a training batch where multiple dataset images are combined using [mosaic augmentation](https://docs.ultralytics.com/reference/data/augment/). Mosaic augmentation increases the diversity of objects and scenes within each batch, helping the model generalize better to various object sizes, aspect ratios, and backgrounds.
This technique is especially valuable for small datasets like COCO8-Multispectral, as it maximizes the utility of each image during training.
## Citations and Acknowledgments
If you use the COCO dataset in your research or development, please cite the following paper:
!!! quote ""
=== "BibTeX"
```bibtex
@misc{lin2015microsoft,
title={Microsoft COCO: Common Objects in Context},
author={Tsung-Yi Lin and Michael Maire and Serge Belongie and Lubomir Bourdev and Ross Girshick and James Hays and Pietro Perona and Deva Ramanan and C. Lawrence Zitnick and Piotr Dollár},
year={2015},
eprint={1405.0312},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
```
Special thanks to the [COCO Consortium](https://cocodataset.org/#home) for their ongoing contributions to the [computer vision community](https://www.ultralytics.com/blog/a-history-of-vision-models).
## FAQ
### What Is the Ultralytics COCO8-Multispectral Dataset Used For?
The Ultralytics COCO8-Multispectral dataset is designed for rapid testing and debugging of [multispectral object detection](https://www.ultralytics.com/glossary/object-detection) models. With only 8 images (4 for training, 4 for validation), it is ideal for verifying your [YOLO26](../../models/yolo26.md) training pipelines and ensuring everything works as expected before scaling to larger datasets. For more datasets to experiment with, visit the [Ultralytics Datasets Catalog](https://docs.ultralytics.com/datasets/).
### How Does Multispectral Data Improve Object Detection?
Multispectral data provides additional spectral information beyond standard RGB, enabling models to distinguish objects based on subtle differences in reflectance across wavelengths. This can enhance detection accuracy, especially in challenging scenarios. Learn more about [multispectral imaging](https://en.wikipedia.org/wiki/Multispectral_imaging) and its applications in [advanced computer vision](https://www.ultralytics.com/blog/ai-in-aviation-a-runway-to-smarter-airports).
### Is COCO8-Multispectral Compatible With Ultralytics Platform and YOLO Models?
Yes, COCO8-Multispectral is fully compatible with [Ultralytics Platform](https://platform.ultralytics.com/) and all [YOLO models](../../models/yolo26.md), including the latest YOLO26. This allows you to easily integrate the dataset into your training and validation workflows.
### Where Can I Find More Information on Data Augmentation Techniques?
For a deeper understanding of data augmentation methods such as mosaic and their impact on model performance, refer to the [YOLO Data Augmentation Guide](https://docs.ultralytics.com/guides/yolo-data-augmentation/) and the [Ultralytics Blog on Data Augmentation](https://www.ultralytics.com/blog/the-ultimate-guide-to-data-augmentation-in-2025).
### Can I Use COCO8-Multispectral for Benchmarking or Educational Purposes?
Absolutely! The small size and multispectral nature of COCO8-Multispectral make it ideal for benchmarking, educational demonstrations, and prototyping new model architectures. For more benchmarking datasets, see the [Ultralytics Benchmark Dataset Collection](https://docs.ultralytics.com/datasets/).

134
algorithms/dms_yolo/code/docs/en/datasets/detect/coco8.md Normal file
View File

@@ -0,0 +1,134 @@
---
comments: true
description: Explore the Ultralytics COCO8 dataset, a versatile and manageable set of 8 images perfect for testing object detection models and training pipelines.
keywords: COCO8, Ultralytics, dataset, object detection, YOLO26, training, validation, machine learning, computer vision
---
# COCO8 Dataset
## Introduction
The [Ultralytics](https://www.ultralytics.com/) COCO8 dataset is a compact yet powerful [object detection](https://www.ultralytics.com/glossary/object-detection) dataset, consisting of the first 8 images from the COCO train 2017 set—4 for training and 4 for validation. This dataset is specifically designed for rapid testing, debugging, and experimentation with [YOLO](https://docs.ultralytics.com/models/yolo26/) models and training pipelines. Its small size makes it highly manageable, while its diversity ensures it serves as an effective sanity check before scaling up to larger datasets.
<p align="center">
<br>
<iframe loading="lazy" width="720" height="405" src="https://www.youtube.com/embed/uDrn9QZJ2lk"
title="YouTube video player" frameborder="0"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen>
</iframe>
<br>
<strong>Watch:</strong> Ultralytics COCO Dataset Overview
</p>
COCO8 is fully compatible with [Ultralytics Platform](https://platform.ultralytics.com/) and [YOLO26](../../models/yolo26.md), enabling seamless integration into your computer vision workflows.
## Dataset YAML
The COCO8 dataset configuration is defined in a YAML (Yet Another Markup Language) file, which specifies dataset paths, class names, and other essential metadata. You can review the official `coco8.yaml` file in the [Ultralytics GitHub repository](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/coco8.yaml).
!!! example "ultralytics/cfg/datasets/coco8.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/coco8.yaml"
```
## Usage
To train a YOLO26n model on the COCO8 dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, use the following examples. For a full list of training options, see the [YOLO Training documentation](../../modes/train.md).
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a pretrained YOLO26n model
model = YOLO("yolo26n.pt")
# Train the model on COCO8
results = model.train(data="coco8.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Train YOLO26n on COCO8 using the command line
yolo detect train data=coco8.yaml model=yolo26n.pt epochs=100 imgsz=640
```
## Sample Images and Annotations
Below is an example of a mosaiced training batch from the COCO8 dataset:
<img src="https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/mosaiced-training-batch-1.avif" alt="COCO8 object detection dataset mosaic training batch" width="800">
- **Mosaiced Image**: This image illustrates a training batch where multiple dataset images are combined using mosaic augmentation. Mosaic augmentation increases the diversity of objects and scenes within each batch, helping the model generalize better to various object sizes, aspect ratios, and backgrounds.
This technique is especially useful for small datasets like COCO8, as it maximizes the value of each image during training.
## Citations and Acknowledgments
If you use the COCO dataset in your research or development, please cite the following paper:
!!! quote ""
=== "BibTeX"
```bibtex
@misc{lin2015microsoft,
title={Microsoft COCO: Common Objects in Context},
author={Tsung-Yi Lin and Michael Maire and Serge Belongie and Lubomir Bourdev and Ross Girshick and James Hays and Pietro Perona and Deva Ramanan and C. Lawrence Zitnick and Piotr Dollár},
year={2015},
eprint={1405.0312},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
```
Special thanks to the [COCO Consortium](https://cocodataset.org/#home) for their ongoing contributions to the [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv) community.
## FAQ
### What Is the Ultralytics COCO8 Dataset Used For?
The Ultralytics COCO8 dataset is designed for rapid testing and debugging of [object detection](https://www.ultralytics.com/glossary/object-detection) models. With only 8 images (4 for training, 4 for validation), it is ideal for verifying your [YOLO](https://docs.ultralytics.com/models/yolo26/) training pipelines and ensuring everything works as expected before scaling to larger datasets. Explore the [COCO8 YAML configuration](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/coco8.yaml) for more details.
### How Do I Train a YOLO26 Model Using the COCO8 Dataset?
You can train a YOLO26 model on COCO8 using either Python or the CLI:
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a pretrained YOLO26n model
model = YOLO("yolo26n.pt")
# Train the model on COCO8
results = model.train(data="coco8.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
yolo detect train data=coco8.yaml model=yolo26n.pt epochs=100 imgsz=640
```
For additional training options, refer to the [YOLO Training documentation](../../modes/train.md).
### Why Should I Use Ultralytics Platform for Managing My COCO8 Training?
[Ultralytics Platform](https://platform.ultralytics.com/) streamlines dataset management, training, and deployment for [YOLO](https://docs.ultralytics.com/models/yolo26/) models—including COCO8. With features like cloud training, real-time monitoring, and intuitive dataset handling, HUB enables you to launch experiments with a single click and eliminates manual setup hassles. Learn more about [Ultralytics Platform](https://platform.ultralytics.com/) and how it can accelerate your computer vision projects.
### What Are the Benefits of Using Mosaic Augmentation in Training With the COCO8 Dataset?
Mosaic augmentation, as used in COCO8 training, combines multiple images into one during each batch. This increases the diversity of objects and backgrounds, helping your [YOLO](https://docs.ultralytics.com/models/yolo26/) model generalize better to new scenarios. Mosaic augmentation is especially valuable for small datasets, as it maximizes the information available in each training step. For more on this, see the [training guide](#usage).
### How Can I Validate My YOLO26 Model Trained on the COCO8 Dataset?
To validate your YOLO26 model after training on COCO8, use the model's validation commands in either Python or CLI. This evaluates your model's performance using standard metrics. For step-by-step instructions, visit the [YOLO Validation documentation](../../modes/val.md).

157
algorithms/dms_yolo/code/docs/en/datasets/detect/construction-ppe.md Normal file
View File

@@ -0,0 +1,157 @@
---
comments: true
description: Discover Construction-PPE, a specialized dataset for detecting helmets, vests, gloves, boots, and goggles in real-world construction sites. Includes compliant and non-compliant scenarios for AI-powered safety monitoring.
keywords: Construction-PPE, PPE dataset, safety compliance, construction workers, object detection, YOLO26, workplace safety, computer vision
---
# Construction-PPE Dataset
<a href="https://colab.research.google.com/github/ultralytics/notebooks/blob/main/notebooks/how-to-train-ultralytics-yolo-on-construction-ppe-detection-dataset.ipynb"><img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Construction-PPE Dataset In Colab"></a>
The Construction-PPE dataset is designed to improve safety compliance in construction sites by enabling detection of essential protective gear such as helmets, vests, gloves, boots, and goggles, along with annotations for missing equipment. Curated from real construction environments, it includes both compliant and non-compliant cases, making it a valuable resource for training AI models that monitor workplace safety.
<p align="center">
<br>
<iframe loading="lazy" width="720" height="405" src="https://www.youtube.com/embed/lFaVnrhMmaE"
title="YouTube video player" frameborder="0"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen>
</iframe>
<br>
<strong>Watch:</strong> How to train Ultralytics YOLO on Personal Protective Equipment Dataset | VisionAI in Construction 👷
</p>
## Dataset Structure
The Construction-PPE dataset is organized into three main subsets:
- **Training Set**: The primary collection of annotated construction images featuring workers with both complete and partial PPE usage.
- **Validation Set**: A designated subset used to fine-tune and assess model performance during PPE detection and compliance monitoring.
- **Test Set**: An independent subset reserved for evaluating the final model's effectiveness in detecting PPE and identifying compliance issues.
Each image is annotated in the [Ultralytics YOLO](../detect/index.md/#what-is-the-ultralytics-yolo-dataset-format-and-how-to-structure-it) format ensuring compatibility with state-of-the-art [object detection](../../tasks/detect.md) and [tracking](../../modes/track.md) pipelines.
The dataset provides **11 classes** divided into positive (worn PPE) and negative (missing PPE) categories. This dual-positive/negative structure enables models to detect properly worn gear **and** identify safety violations.
## Business Value
- Construction remains one of the most hazardous industries in the world, with over 51 out of 123 work related **fatal injuries** in the UK in 2023/2024 happening in construction. However, the issue is no longer an issue with lack of regulation with 42% of construction workers admitting to not always adhering to processes.
- Construction is already governed by an extensive framework of health and safety (HSE) standards, but HSE teams are challenged with consistent enforcement. HSE teams are often stretched thin, balancing paperwork and audits and lacking the ability to monitor every corner of a busy and ever-changing environment in real time.
- This is where computer vision based personal protective equipment (PPE) detection becomes invaluable. By automatically checking whether workers are wearing **helmets, vests and other personal protective equipment**, you can ensure HSE rules are not just present but effectively enforced consistently across all sites. Beyond compliance, computer vision provides leading indicators of risk by revealing how well crews follow safety practices, enabling organizations to spot downward trends in compliance and prevent incidents before they happen.
- As a bonus, personal protective equipment detection has also been known to identify unauthorized site intruders, since **those not equipped with proper safety gear** are the first to trigger a notification. Ultimately, PPE detection is a simple yet powerful computer vision use-case that delivers full oversight, actionable insights and standardized reporting, empowering construction firms to reduce risk, protect workers and safeguard their projects.
## Applications
Construction-PPE powers a variety of safety-focused computer vision applications:
- **Automated compliance monitoring**: Train AI models to instantly check if workers are wearing required safety gear like helmets, vests, or gloves, reducing risks on site.
- **Workplace safety analytics**: Track PPE usage over time, spot frequent violations, and generate insights to improve safety culture.
- **Smart surveillance systems**: Connect detection models with cameras to send real-time alerts when PPE is missing, preventing accidents before they happen.
- **Robotics and autonomous systems**: Enable drones or robots to perform PPE checks across large sites, supporting faster and safer inspections.
- **Research and education**: Provide a real-world dataset for students and researchers exploring workplace safety and human-object interactions.
## Dataset YAML
The Construction-PPE dataset includes a YAML configuration file that defines the training and validation image paths along with the full list of object classes. You can access the `construction-ppe.yaml` file directly in the Ultralytics repository here: [https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/construction-ppe.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/construction-ppe.yaml)
!!! example "ultralytics/cfg/datasets/construction-ppe.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/construction-ppe.yaml"
```
## Usage
You can train a YOLO26n model on the Construction-PPE dataset for 100 epochs with an image size of 640. The following examples show how to get started quickly. For more options and advanced configurations, see the [Training guide](../../modes/train.md).
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load pretrained model
model = YOLO("yolo26n.pt")
# Train the model on Construction-PPE dataset
model.train(data="construction-ppe.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
yolo detect train data=construction-ppe.yaml model=yolo26n.pt epochs=100 imgsz=640
```
## Sample Images and Annotations
The dataset captures construction workers across varied environments, lighting conditions, and postures. Both **compliant** and **non-compliant** cases are included.
![Construction-PPE dataset sample with safety gear detection](https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/construction-ppe-dataset-sample.avif)
## License and Attribution
Construction-PPE is developed and released under the [AGPL-3.0 License](https://github.com/ultralytics/ultralytics/blob/main/LICENSE), supporting open-source research and commercial applications with proper attribution.
If you use this dataset in your research, please cite it:
!!! quote ""
=== "BibTeX"
```bibtex
@dataset{Dalvi_Construction_PPE_Dataset_2025,
author = {Mrunmayee Dalvi and Niyati Singh and Sahil Bhingarde and Ketaki Chalke},
title = {Construction-PPE: Personal Protective Equipment Detection Dataset},
month = {January},
year = {2025},
version = {1.0.0},
license = {AGPL-3.0},
url = {https://docs.ultralytics.com/datasets/detect/construction-ppe/},
publisher = {Ultralytics}
}
```
## FAQ
### What makes the Construction-PPE dataset unique?
Unlike generic construction datasets, Construction-PPE explicitly includes **missing equipment classes**. This dual-labeling approach allows models to not only detect PPE but also flag violations in real-time.
### Which object categories are included?
The dataset covers helmets, vests, gloves, boots, goggles, and workers, along with their “missing PPE” counterparts. This ensures comprehensive compliance coverage.
### How can I train a YOLO model using the Construction-PPE dataset?
To train a YOLO26 model using the Construction-PPE dataset, you can use the following code snippets:
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="construction-ppe.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=construction-ppe.yaml model=yolo26n.pt epochs=100 imgsz=640
```
### Is this dataset suitable for real-world applications?
Yes. Images are curated from real construction sites under diverse conditions. This makes it highly effective for building deployable workplace safety monitoring systems.
### What are the benefits of using the Construction-PPE dataset in AI projects?
The dataset enables real-time detection of personal protective equipment, helping monitor worker safety on construction sites. With classes for both worn and missing gear, it supports AI systems that can automatically flag safety violations, generate compliance insights, and reduce risks. It also provides a practical resource for developing computer vision solutions in workplace safety, robotics, and academic research.

145
algorithms/dms_yolo/code/docs/en/datasets/detect/globalwheat2020.md Normal file
View File

@@ -0,0 +1,145 @@
---
comments: true
description: Explore the Global Wheat Head Dataset to develop accurate wheat head detection models. Includes training images, annotations, and usage for crop management.
keywords: Global Wheat Head Dataset, wheat head detection, wheat phenotyping, crop management, deep learning, object detection, training datasets
---
# Global Wheat Head Dataset
The [Global Wheat Head Dataset](https://www.global-wheat.com/) is a collection of images designed to support the development of accurate wheat head detection models for applications in wheat phenotyping and crop management. Wheat heads, also known as spikes, are the grain-bearing parts of the wheat plant. Accurate estimation of wheat head density and size is essential for assessing crop health, maturity, and yield potential. The dataset, created by a collaboration of nine research institutes from seven countries, covers multiple growing regions to ensure models generalize well across different environments.
## Key Features
- The dataset contains over 3,000 training images from Europe (France, UK, Switzerland) and North America (Canada).
- It includes approximately 1,000 test images from Australia, Japan, and China.
- Images are outdoor field images, capturing the natural variability in wheat head appearances.
- Annotations include wheat head bounding boxes to support [object detection](https://docs.ultralytics.com/tasks/detect/) tasks.
## Dataset Structure
The Global Wheat Head Dataset is organized into two main subsets:
1. **Training Set**: This subset contains over 3,000 images from Europe and North America. The images are labeled with wheat head bounding boxes, providing ground truth for training object detection models.
2. **Test Set**: This subset consists of approximately 1,000 images from Australia, Japan, and China. These images are used for evaluating the performance of trained models on unseen genotypes, environments, and observational conditions.
## Applications
The Global Wheat Head Dataset is widely used for training and evaluating [deep learning](https://www.ultralytics.com/glossary/deep-learning-dl) models in wheat head detection tasks. The dataset's diverse set of images, capturing a wide range of appearances, environments, and conditions, make it a valuable resource for researchers and practitioners in the field of [plant phenotyping](https://www.ultralytics.com/blog/computer-vision-in-agriculture-transforming-fruit-detection-and-precision-farming) and crop management.
## Dataset YAML
A YAML (Yet Another Markup Language) file is used to define the dataset configuration. It contains information about the dataset's paths, classes, and other relevant information. For the case of the Global Wheat Head Dataset, the `GlobalWheat2020.yaml` file is maintained at [https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/GlobalWheat2020.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/GlobalWheat2020.yaml).
!!! example "ultralytics/cfg/datasets/GlobalWheat2020.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/GlobalWheat2020.yaml"
```
## Usage
To train a YOLO26n model on the Global Wheat Head Dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, you can use the following code snippets. For a comprehensive list of available arguments, refer to the model [Training](../../modes/train.md) page.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="GlobalWheat2020.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=GlobalWheat2020.yaml model=yolo26n.pt epochs=100 imgsz=640
```
## Sample Data and Annotations
The Global Wheat Head Dataset contains a diverse set of outdoor field images, capturing the natural variability in wheat head appearances, environments, and conditions. Here are some examples of data from the dataset, along with their corresponding annotations:
![Global Wheat dataset sample showing wheat head detection](https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/wheat-head-detection-sample.avif)
- **Wheat Head Detection**: This image demonstrates an example of wheat head detection, where wheat heads are annotated with bounding boxes. The dataset provides a variety of images to facilitate the development of models for this task.
The example showcases the variety and complexity of the data in the Global Wheat Head Dataset and highlights the importance of accurate wheat head detection for applications in wheat phenotyping and crop management.
## Citations and Acknowledgments
If you use the Global Wheat Head Dataset in your research or development work, please cite the following paper:
!!! quote ""
=== "BibTeX"
```bibtex
@article{david2020global,
title={Global Wheat Head Detection (GWHD) Dataset: A Large and Diverse Dataset of High-Resolution RGB-Labelled Images to Develop and Benchmark Wheat Head Detection Methods},
author={David, Etienne and Madec, Simon and Sadeghi-Tehran, Pouria and Aasen, Helge and Zheng, Bangyou and Liu, Shouyang and Kirchgessner, Norbert and Ishikawa, Goro and Nagasawa, Koichi and Badhon, Minhajul and others},
journal={arXiv preprint arXiv:2005.02162},
year={2020}
}
```
We would like to acknowledge the researchers and institutions that contributed to the creation and maintenance of the Global Wheat Head Dataset as a valuable resource for the plant phenotyping and crop management research community. For more information about the dataset and its creators, visit the [Global Wheat Head Dataset website](https://www.global-wheat.com/).
## FAQ
### What is the Global Wheat Head Dataset used for?
The Global Wheat Head Dataset is primarily used for developing and training deep learning models aimed at wheat head detection. This is crucial for applications in [wheat phenotyping](https://www.ultralytics.com/blog/from-farm-to-table-how-ai-drives-innovation-in-agriculture) and crop management, allowing for more accurate estimations of wheat head density, size, and overall crop yield potential. Accurate detection methods help in assessing crop health and maturity, essential for efficient crop management.
### How do I train a YOLO26n model on the Global Wheat Head Dataset?
To train a YOLO26n model on the Global Wheat Head Dataset, you can use the following code snippets. Make sure you have the `GlobalWheat2020.yaml` configuration file specifying dataset paths and classes:
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a pretrained model (recommended for training)
model = YOLO("yolo26n.pt")
# Train the model
results = model.train(data="GlobalWheat2020.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=GlobalWheat2020.yaml model=yolo26n.pt epochs=100 imgsz=640
```
For a comprehensive list of available arguments, refer to the model [Training](../../modes/train.md) page.
### What are the key features of the Global Wheat Head Dataset?
Key features of the Global Wheat Head Dataset include:
- Over 3,000 training images from Europe (France, UK, Switzerland) and North America (Canada).
- Approximately 1,000 test images from Australia, Japan, and China.
- High variability in wheat head appearances due to different growing environments.
- Detailed annotations with wheat head bounding boxes to aid [object detection](https://www.ultralytics.com/glossary/object-detection) models.
These features facilitate the development of robust models capable of generalization across multiple regions.
### Where can I find the configuration YAML file for the Global Wheat Head Dataset?
The configuration YAML file for the Global Wheat Head Dataset, named `GlobalWheat2020.yaml`, is available on GitHub. You can access it at <https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/GlobalWheat2020.yaml>. This file contains necessary information about dataset paths, classes, and other configuration details needed for model training in [Ultralytics YOLO](https://docs.ultralytics.com/models/yolo26/).
### Why is wheat head detection important in crop management?
Wheat head detection is critical in crop management because it enables accurate estimation of wheat head density and size, which are essential for evaluating crop health, maturity, and yield potential. By leveraging [deep learning models](https://docs.ultralytics.com/models/) trained on datasets like the Global Wheat Head Dataset, farmers and researchers can better monitor and manage crops, leading to improved productivity and optimized resource use in agricultural practices. This technological advancement supports [sustainable agriculture](https://www.ultralytics.com/blog/real-time-crop-health-monitoring-with-ultralytics-yolo11) and food security initiatives.
For more information on applications of AI in agriculture, visit [AI in Agriculture](https://www.ultralytics.com/solutions/ai-in-agriculture).

169
algorithms/dms_yolo/code/docs/en/datasets/detect/homeobjects-3k.md Normal file
View File

@@ -0,0 +1,169 @@
---
comments: true
description: Discover HomeObjects-3K, a rich indoor object detection dataset with 12 classes like bed, sofa, TV, and laptop. Ideal for computer vision in smart homes, robotics, and AR.
keywords: HomeObjects-3K, indoor dataset, household items, object detection, computer vision, YOLO26, smart home AI, robotics dataset
---
# HomeObjects-3K Dataset
<a href="https://colab.research.google.com/github/ultralytics/notebooks/blob/main/notebooks/how-to-train-ultralytics-yolo-on-homeobjects-dataset.ipynb"><img src="https://colab.research.google.com/assets/colab-badge.svg" alt="HomeObjects-3K Dataset In Colab"></a>
The HomeObjects-3K dataset is a curated collection of common household object images, designed for training, testing, and [benchmarking](../../modes/benchmark.md) [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv) models. Featuring ~3,000 images and 12 distinct object classes, this dataset is ideal for research and applications in indoor scene understanding, smart home devices, [robotics](https://www.ultralytics.com/glossary/robotics), and augmented reality.
<p align="center">
<br>
<iframe loading="lazy" width="720" height="405" src="https://www.youtube.com/embed/v3iqOYoRBFQ"
title="YouTube video player" frameborder="0"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen>
</iframe>
<br>
<strong>Watch:</strong> How to Train Ultralytics YOLO26 on HomeObjects-3K Dataset | Detection, Validation & ONNX Export 🚀
</p>
## Dataset Structure
The HomeObjects-3K dataset is organized into the following subsets:
- **Training Set**: Comprises 2,285 annotated images featuring objects such as sofas, chairs, tables, lamps, and more.
- **Validation Set**: Includes 404 annotated images designated for evaluating model performance.
Each image is labeled using bounding boxes aligned with the [Ultralytics YOLO](../detect/index.md/#what-is-the-ultralytics-yolo-dataset-format-and-how-to-structure-it) format. The diversity of indoor lighting, object scale, and orientations makes it robust for real-world deployment scenarios.
## Object Classes
The dataset supports 12 everyday object categories, covering furniture, electronics, and decorative items. These classes are chosen to reflect common items encountered in indoor domestic environments and support vision tasks like [object detection](../../tasks/detect.md) and [object tracking](../../modes/track.md).
!!! Tip "HomeObjects-3K classes"
0. bed
1. sofa
2. chair
3. table
4. lamp
5. tv
6. laptop
7. wardrobe
8. window
9. door
10. potted plant
11. photo frame
## Applications
HomeObjects-3K enables a wide spectrum of applications in indoor computer vision, spanning both research and real-world product development:
- **Indoor object detection**: Use models like [Ultralytics YOLO26](../../models/yolo26.md) to find and locate common home items like beds, chairs, lamps, and laptops in images. This helps with real-time understanding of indoor scenes.
- **Scene layout parsing**: In robotics and smart home systems, this helps devices understand how rooms are arranged, where objects like doors, windows, and furniture are, so they can navigate safely and interact with their environment properly.
- **AR applications**: Power [object recognition](http://ultralytics.com/glossary/image-recognition) features in apps that use augmented reality. For example, detect TVs or wardrobes and show extra information or effects on them.
- **Education and research**: Support learning and academic projects by giving students and researchers a ready-to-use dataset for practicing indoor object detection with real-world examples.
- **Home inventory and asset tracking**: Automatically detect and list home items in photos or videos, useful for managing belongings, organizing spaces, or visualizing furniture in real estate.
## Dataset YAML
The configuration for the HomeObjects-3K dataset is provided through a YAML file. This file outlines essential information such as image paths for train and validation directories, and the list of object classes.
You can access the `HomeObjects-3K.yaml` file directly from the Ultralytics repository at: [https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/HomeObjects-3K.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/HomeObjects-3K.yaml)
!!! example "ultralytics/cfg/datasets/HomeObjects-3K.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/HomeObjects-3K.yaml"
```
## Usage
You can train a YOLO26n model on the HomeObjects-3K dataset for 100 epochs using an image size of 640. The examples below show how to get started. For more training options and detailed settings, check the [Training](../../modes/train.md) guide.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load pretrained model
model = YOLO("yolo26n.pt")
# Train the model on HomeObjects-3K dataset
model.train(data="HomeObjects-3K.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
yolo detect train data=HomeObjects-3K.yaml model=yolo26n.pt epochs=100 imgsz=640
```
## Sample Images and Annotations
The dataset features a rich collection of indoor scene images that capture a wide range of household objects in natural home environments. Below are sample visuals from the dataset, each paired with its corresponding annotations to illustrate object positions, scales, and spatial relationships.
![HomeObjects-3K dataset sample with household objects](https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/homeobjects-3k-dataset-sample.avif)
## License and Attribution
HomeObjects-3K is developed and released by the **[Ultralytics team](https://www.ultralytics.com/about)** under the [AGPL-3.0 License](https://github.com/ultralytics/ultralytics/blob/main/LICENSE), supporting open-source research and commercial use with proper attribution.
If you use this dataset in your research, please cite it using the mentioned details:
!!! quote ""
=== "BibTeX"
```bibtex
@dataset{Jocher_Ultralytics_Datasets_2025,
author = {Jocher, Glenn and Rizwan, Muhammad},
license = {AGPL-3.0},
month = {May},
title = {Ultralytics Datasets: HomeObjects-3K Detection Dataset},
url = {https://docs.ultralytics.com/datasets/detect/homeobjects-3k/},
version = {1.0.0},
year = {2025}
}
```
## FAQ
### What is the HomeObjects-3K dataset designed for?
HomeObjects-3K is crafted for advancing AI understanding of indoor scenes. It focuses on detecting everyday household items—like beds, sofas, TVs, and lamps—making it ideal for applications in smart homes, robotics, augmented reality, and interior monitoring systems. Whether you're training models for real-time edge devices or academic research, this dataset provides a balanced foundation.
### Which object categories are included, and why were they selected?
The dataset includes 12 of the most commonly encountered household items: bed, sofa, chair, table, lamp, tv, laptop, wardrobe, window, door, potted plant, and photo frame. These objects were chosen to reflect realistic indoor environments and to support multipurpose tasks such as robotic navigation, or scene generation in AR/VR applications.
### How can I train a YOLO model using the HomeObjects-3K dataset?
To train a YOLO model like YOLO26n, you'll just need the `HomeObjects-3K.yaml` configuration file and the [pretrained model](../../models/index.md) weights. Whether you're using Python or the CLI, training can be launched with a single command. You can customize parameters such as epochs, image size, and batch size depending on your target performance and hardware setup.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load pretrained model
model = YOLO("yolo26n.pt")
# Train the model on HomeObjects-3K dataset
model.train(data="HomeObjects-3K.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
yolo detect train data=HomeObjects-3K.yaml model=yolo26n.pt epochs=100 imgsz=640
```
### Is this dataset suitable for beginner-level projects?
Absolutely. With clean labeling, and standardized YOLO-compatible annotations, HomeObjects-3K is an excellent entry point for students and hobbyists who want to explore real-world object detection in indoor scenarios. It also scales well for more complex applications in commercial environments.
### Where can I find the annotation format and YAML?
Refer to the [Dataset YAML](#dataset-yaml) section. The format is standard YOLO, making it compatible with most object detection pipelines.

343
algorithms/dms_yolo/code/docs/en/datasets/detect/index.md Normal file
View File

@@ -0,0 +1,343 @@
---
comments: true
description: Learn about dataset formats compatible with Ultralytics YOLO for robust object detection. Explore supported datasets and learn how to convert formats.
keywords: Ultralytics, YOLO, object detection datasets, dataset formats, COCO, dataset conversion, training datasets
---
# Object Detection Datasets Overview
Training a robust and accurate [object detection](https://www.ultralytics.com/glossary/object-detection) model requires a comprehensive dataset. This guide introduces various formats of datasets that are compatible with the Ultralytics YOLO model and provides insights into their structure, usage, and how to convert between different formats.
## Supported Dataset Formats
### Ultralytics YOLO format
The Ultralytics YOLO format is a dataset configuration format that allows you to define the dataset root directory, the relative paths to training/validation/testing image directories or `*.txt` files containing image paths, and a dictionary of class names. Here is an example:
!!! example "ultralytics/cfg/datasets/coco8.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/coco8.yaml"
```
Labels for this format should be exported to YOLO format with one `*.txt` file per image. If there are no objects in an image, no `*.txt` file is required. The `*.txt` file should be formatted with one row per object in `class x_center y_center width height` format. Box coordinates must be in **normalized xywh** format (from 0 to 1). If your boxes are in pixels, you should divide `x_center` and `width` by image width, and `y_center` and `height` by image height. Class numbers should be zero-indexed (start with 0).
<p align="center"><img width="750" src="https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/two-persons-tie.avif" alt="YOLO labeled image with bounding boxes on persons and tie"></p>
The label file corresponding to the above image contains 2 persons (class `0`) and a tie (class `27`):
<p align="center"><img width="428" src="https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/two-persons-tie-1.avif" alt="YOLO format label file with normalized coordinates"></p>
When using the Ultralytics YOLO format, organize your training and validation images and labels as shown in the [COCO8 dataset](coco8.md) example below.
<p align="center"><img width="800" src="https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/two-persons-tie-2.avif" alt="YOLO dataset directory structure with train and val folders"></p>
#### Usage Example
Here's how you can use YOLO format datasets to train your model:
!!! example
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="coco8.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=coco8.yaml model=yolo26n.pt epochs=100 imgsz=640
```
### Ultralytics NDJSON format
The NDJSON (Newline Delimited JSON) format provides an alternative way to define datasets for Ultralytics YOLO models. This format stores dataset metadata and annotations in a single file where each line contains a separate JSON object.
An NDJSON dataset file contains:
1. **Dataset record** (first line): Contains dataset metadata including task type, class names, and general information
2. **Image records** (subsequent lines): Contains individual image data including dimensions, annotations, and file paths
!!! example "NDJSON Example"
=== "Dataset record (line 1)"
```json
{
"type": "dataset",
"task": "detect",
"name": "Example",
"description": "COCO NDJSON example dataset",
"url": "https://app.ultralytics.com/user/datasets/example",
"class_names": { "0": "person", "1": "bicycle", "2": "car" },
"bytes": 426342,
"version": 0,
"created_at": "2024-01-01T00:00:00Z",
"updated_at": "2025-01-01T00:00:00Z"
}
```
=== "Detect"
```json
{
"type": "image",
"file": "image1.jpg",
"url": "https://www.url.com/path/to/image1.jpg",
"width": 640,
"height": 480,
"split": "train",
"annotations": {
"boxes": [
[0, 0.525, 0.376, 0.284, 0.418],
[1, 0.735, 0.298, 0.193, 0.337]
]
}
}
```
Format: `[class_id, x_center, y_center, width, height]`
=== "Segment"
```json
{
"type": "image",
"file": "image1.jpg",
"url": "https://www.url.com/path/to/image1.jpg",
"width": 640,
"height": 480,
"split": "train",
"annotations": {
"segments": [
[0, 0.681, 0.485, 0.670, 0.487, 0.676, 0.487, 0.688, 0.515],
[1, 0.422, 0.315, 0.438, 0.330, 0.445, 0.328, 0.450, 0.320]
]
}
}
```
Format: `[class_id, x1, y1, x2, y2, x3, y3, ...]`
=== "Pose"
```json
{
"type": "image",
"file": "image1.jpg",
"url": "https://www.url.com/path/to/image1.jpg",
"width": 640,
"height": 480,
"split": "train",
"annotations": {
"pose": [
[0, 0.523, 0.376, 0.283, 0.418, 0.374, 0.169, 2, 0.364, 0.178, 2],
[0, 0.735, 0.298, 0.193, 0.337, 0.412, 0.225, 2, 0.408, 0.231, 2]
]
}
}
```
Format: `[class_id, x_center, y_center, width, height, x1, y1, v1, x2, y2, v2, ...]`
Keypoints follow bbox as repeated `(x, y, v)` triplets where `v` is visibility: 0=not labeled, 1=labeled but occluded, 2=labeled and visible. The keypoint count is dataset-specific (e.g., COCO pose has 17 keypoints = 51 values after bbox).
=== "OBB"
```json
{
"type": "image",
"file": "image1.jpg",
"url": "https://www.url.com/path/to/image1.jpg",
"width": 640,
"height": 480,
"split": "train",
"annotations": {
"obb": [
[0, 0.480, 0.352, 0.568, 0.356, 0.572, 0.400, 0.484, 0.396],
[1, 0.711, 0.274, 0.759, 0.278, 0.755, 0.322, 0.707, 0.318]
]
}
}
```
Format: `[class_id, x1, y1, x2, y2, x3, y3, x4, y4]`
The four corner points define the oriented bounding box in clockwise order starting from the top-left corner. All coordinates are normalized (0-1).
=== "Classify"
```json
{
"type": "image",
"file": "image1.jpg",
"url": "https://www.url.com/path/to/image1.jpg",
"width": 640,
"height": 480,
"split": "train",
"annotations": {
"classification": [0]
}
}
```
Format: `[class_id]`
#### Usage Example
To use an NDJSON dataset with YOLO26, simply specify the path to the `.ndjson` file:
!!! example
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt")
# Train using NDJSON dataset
results = model.train(data="path/to/dataset.ndjson", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training with NDJSON dataset
yolo detect train data=path/to/dataset.ndjson model=yolo26n.pt epochs=100 imgsz=640
```
#### Advantages of NDJSON format
- **Single file**: All dataset information contained in one file
- **Streaming**: Can process large datasets line-by-line without loading everything into memory
- **Cloud integration**: Supports remote image URLs for cloud-based training
- **Extensible**: Easy to add custom metadata fields
- **Version control**: Single file format works well with git and version control systems
## Supported Datasets
Here is a list of the supported datasets and a brief description for each:
- [African-wildlife](african-wildlife.md): A dataset featuring images of African wildlife, including buffalo, elephant, rhino, and zebras.
- [Argoverse](argoverse.md): A dataset containing 3D tracking and motion forecasting data from urban environments with rich annotations.
- [Brain-tumor](brain-tumor.md): A dataset for detecting brain tumors includes MRI or CT scan images with details on tumor presence, location, and characteristics.
- [COCO](coco.md): Common Objects in Context (COCO) is a large-scale [object detection](https://www.ultralytics.com/glossary/object-detection), segmentation, and captioning dataset with 80 object categories.
- [COCO8](coco8.md): A smaller subset of the first 4 images from COCO train and COCO val, suitable for quick tests.
- [COCO8-Grayscale](coco8-grayscale.md): A grayscale version of COCO8 created by converting RGB to grayscale, useful for single-channel model evaluation.
- [COCO8-Multispectral](coco8-multispectral.md): A 10-channel multispectral version of COCO8 created by interpolating RGB wavelengths, useful for spectral-aware model evaluation.
- [COCO12-Formats](coco12-formats.md): A test dataset with 12 images covering all supported image formats (AVIF, BMP, DNG, HEIC, JP2, JPEG, JPG, MPO, PNG, TIF, TIFF, WebP) for validating image loading pipelines.
- [COCO128](coco128.md): A smaller subset of the first 128 images from COCO train and COCO val, suitable for tests.
- [Construction-PPE](construction-ppe.md): A dataset featuring construction site workers with labeled safety gear such as helmets, vests, gloves, boots, and goggles, including missing-equipment annotations like no_helmet, no_googles for real-world compliance monitoring.
- [Global Wheat 2020](globalwheat2020.md): A dataset containing images of wheat heads for the Global Wheat Challenge 2020.
- [HomeObjects-3K](homeobjects-3k.md): A dataset of indoor household items including beds, chairs, TVs, and more—ideal for applications in smart home automation, robotics, augmented reality, and room layout analysis.
- [KITTI](kitti.md): A dataset featuring real-world driving scenes with stereo, LiDAR, and GPS/IMU data, used here for **2D object detection** tasks such as identifying cars, pedestrians, and cyclists in urban, rural, and highway environments.
- [LVIS](lvis.md): A large-scale object detection, segmentation, and captioning dataset with 1203 object categories.
- [Medical-pills](medical-pills.md): A dataset featuring images of medical-pills, annotated for applications such as pharmaceutical quality assurance, pill sorting, and regulatory compliance.
- [Objects365](objects365.md): A high-quality, large-scale dataset for object detection with 365 object categories and over 600K annotated images.
- [OpenImagesV7](open-images-v7.md): A comprehensive dataset by Google with 1.7M train images and 42k validation images.
- [Roboflow 100](roboflow-100.md): A diverse object detection benchmark with 100 datasets spanning seven imagery domains for comprehensive model evaluation.
- [Signature](signature.md): A dataset featuring images of various documents with annotated signatures, supporting document verification and fraud detection research.
- [SKU-110K](sku-110k.md): A dataset featuring dense object detection in retail environments with over 11K images and 1.7 million [bounding boxes](https://www.ultralytics.com/glossary/bounding-box).
- [TT100K](tt100k.md): Explore the Tsinghua-Tencent 100K (TT100K) traffic sign dataset with 100,000 street view images and 30,000+ annotated traffic signs for robust detection and classification.
- [VisDrone](visdrone.md): A dataset containing object detection and multi-object tracking data from drone-captured imagery with over 10K images and video sequences.
- [VOC](voc.md): The Pascal Visual Object Classes (VOC) dataset for object detection and segmentation with 20 object classes and over 11K images.
- [xView](xview.md): A dataset for object detection in overhead imagery with 60 object categories and over 1 million annotated objects.
### Adding your own dataset
If you have your own dataset and would like to use it for training detection models with Ultralytics YOLO format, ensure that it follows the format specified above under "Ultralytics YOLO format". Convert your annotations to the required format and specify the paths, number of classes, and class names in the YAML configuration file.
## Port or Convert Label Formats
### COCO Dataset Format to YOLO Format
You can easily convert labels from the popular [COCO dataset](coco.md) format to the YOLO format using the following code snippet:
!!! example
=== "Python"
```python
from ultralytics.data.converter import convert_coco
convert_coco(labels_dir="path/to/coco/annotations/")
```
This conversion tool can be used to convert the COCO dataset or any dataset in the COCO format to the Ultralytics YOLO format. The process transforms the JSON-based COCO annotations into the simpler text-based YOLO format, making it compatible with [Ultralytics YOLO models](../../models/yolo26.md).
Remember to double-check if the dataset you want to use is compatible with your model and follows the necessary format conventions. Properly formatted datasets are crucial for training successful object detection models.
## FAQ
### What is the Ultralytics YOLO dataset format and how to structure it?
The Ultralytics YOLO format is a structured configuration for defining datasets in your training projects. It involves setting paths to your training, validation, and testing images and corresponding labels. For example:
```yaml
--8<-- "ultralytics/cfg/datasets/coco8.yaml"
```
Labels are saved in `*.txt` files with one file per image, formatted as `class x_center y_center width height` with normalized coordinates. For a detailed guide, see the [COCO8 dataset example](coco8.md).
### How do I convert a COCO dataset to the YOLO format?
You can convert a COCO dataset to the YOLO format using the [Ultralytics conversion tools](../../reference/data/converter.md). Here's a quick method:
```python
from ultralytics.data.converter import convert_coco
convert_coco(labels_dir="path/to/coco/annotations/")
```
This code will convert your COCO annotations to YOLO format, enabling seamless integration with Ultralytics YOLO models. For additional details, visit the [Port or Convert Label Formats](#port-or-convert-label-formats) section.
### Which datasets are supported by Ultralytics YOLO for object detection?
Ultralytics YOLO supports a wide range of datasets, including:
- [Argoverse](argoverse.md)
- [COCO](coco.md)
- [LVIS](lvis.md)
- [COCO8](coco8.md)
- [Global Wheat 2020](globalwheat2020.md)
- [Objects365](objects365.md)
- [OpenImagesV7](open-images-v7.md)
Each dataset page provides detailed information on the structure and usage tailored for efficient YOLO26 training. Explore the full list in the [Supported Datasets](#supported-datasets) section.
### How do I start training a YOLO26 model using my dataset?
To start training a YOLO26 model, ensure your dataset is formatted correctly and the paths are defined in a YAML file. Use the following script to begin training:
!!! example
=== "Python"
```python
from ultralytics import YOLO
model = YOLO("yolo26n.pt") # Load a pretrained model
results = model.train(data="path/to/your_dataset.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
yolo detect train data=path/to/your_dataset.yaml model=yolo26n.pt epochs=100 imgsz=640
```
Refer to the [Usage](#usage-example) section for more details on utilizing different modes, including CLI commands.
### Where can I find practical examples of using Ultralytics YOLO for object detection?
Ultralytics provides numerous examples and practical guides for using YOLO26 in diverse applications. For a comprehensive overview, visit the [Ultralytics Blog](https://www.ultralytics.com/blog) where you can find case studies, detailed tutorials, and community stories showcasing object detection, segmentation, and more with YOLO26. For specific examples, check the [Usage](../../modes/predict.md) section in the documentation.

127
algorithms/dms_yolo/code/docs/en/datasets/detect/kitti.md Normal file
View File

@@ -0,0 +1,127 @@
---
comments: true
description: Explore the Ultralytics kitti dataset, a benchmark dataset for computer vision tasks such as 3D object detection, depth estimation, and autonomous driving perception.
keywords: kitti, Ultralytics, dataset, object detection, 3D vision, YOLO26, training, validation, self-driving cars, computer vision
---
# KITTI Dataset
<a href="https://colab.research.google.com/github/ultralytics/notebooks/blob/main/notebooks/how-to-train-ultralytics-yolo-on-kitti-detection-dataset.ipynb"><img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open KITTI Dataset In Colab"></a>
The kitti dataset is one of the most influential benchmark datasets for autonomous driving and computer vision. Released by the Karlsruhe Institute of Technology and Toyota Technological Institute at Chicago, it contains stereo camera, LiDAR, and GPS/IMU data collected from real-world driving scenarios.
<p align="center">
<br>
<iframe loading="lazy" width="720" height="405" src="https://www.youtube.com/embed/NNeDlTbq9pA"
title="YouTube video player" frameborder="0"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen>
</iframe>
<br>
<strong>Watch:</strong> How to Train Ultralytics YOLO26 on the KITTI Dataset 🚀
</p>
It is widely used for evaluating algorithms in object detection, depth estimation, optical flow, and visual odometry. The dataset is fully compatible with Ultralytics YOLO26 for 2D object detection tasks and can be easily integrated into the Ultralytics platform for training and evaluation.
## Dataset Structure
!!! warning
Kitti original test set is excluded here since it does not contain ground-truth annotations.
In total, the dataset includes 7,481 images, each paired with detailed annotations for objects such as cars, pedestrians, cyclists, and other road elements. The dataset is divided into two main subsets:
- **Training set:** Contains 5,985 images with annotated labels used for model training.
- **Validation set:** Includes 1,496 images with corresponding annotations used for performance evaluation and benchmarking.
## Applications
Kitti dataset enables advancements in autonomous driving and robotics, supporting tasks like:
- **Autonomous vehicle perception**: Training models to detect and track vehicles, pedestrians, and obstacles for safe navigation in self-driving systems.
- **3D scene understanding**: Supporting depth estimation, stereo vision, and 3D object localization to help machines understand spatial environments.
- **Optical flow and motion prediction**: Enabling motion analysis to predict the movement of objects and improve trajectory planning in dynamic environments.
- **Computer vision benchmarking**: Serving as a standard benchmark for evaluating performance across multiple vision tasks, including object detection, and tracking.
## Dataset YAML
Ultralytics defines the kitti dataset configuration using a YAML file. This file specifies dataset paths, class labels, and metadata required for training. The configuration file is available at [https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/kitti.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/kitti.yaml).
!!! example "ultralytics/cfg/datasets/kitti.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/kitti.yaml"
```
## Usage
To train a YOLO26n model on the kitti dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, use the following commands. For more details, refer to the [Training](../../modes/train.md) page.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a pretrained YOLO26 model
model = YOLO("yolo26n.pt")
# Train on kitti dataset
results = model.train(data="kitti.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
yolo detect train data=kitti.yaml model=yolo26n.pt epochs=100 imgsz=640
```
You can also perform evaluation, [inference](../../modes/predict.md), and [export](../../modes/export.md) tasks directly from the command line or Python API using the same configuration file.
## Sample Images and Annotations
The kitti dataset provides diverse driving scenarios. Each image includes bounding box annotations for 2D object detection tasks. The example showcase the dataset rich variety, enabling robust model generalization across diverse real-world conditions.
<img src="https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/kitti-dataset-sample.avif" alt="KITTI dataset vehicle detection sample" width="800">
## Citations and Acknowledgments
If you use the kitti dataset in your research, please cite the following paper:
!!! quote
=== "BibTeX"
```bibtex
@article{Geiger2013IJRR,
author = {Andreas Geiger and Philip Lenz and Christoph Stiller and Raquel Urtasun},
title = {Vision meets Robotics: The KITTI Dataset},
journal = {International Journal of Robotics Research (IJRR)},
year = {2013}
}
```
We acknowledge the KITTI Vision Benchmark Suite for providing this comprehensive dataset that continues to shape progress in computer vision, robotics, and autonomous systems. Visit the [kitti website](https://www.cvlibs.net/datasets/kitti/) for more information.
## FAQs
### What is the kitti dataset used for?
The kitti dataset is primarily used for computer vision research in autonomous driving, supporting tasks like object detection, depth estimation, optical flow, and 3D localization.
### How many images are included in the kitti dataset?
The dataset includes 5,985 labeled training images and 1,496 validation images captured across urban, rural, and highway scenes. The original test set is excluded here since it does not contain ground-truth annotations.
### Which object classes are annotated in the dataset?
kitti includes annotations for objects such as cars, pedestrians, cyclists, trucks, trams, and miscellaneous road users.
### Can I train Ultralytics YOLO26 models using the kitti dataset?
Yes, kitti is fully compatible with Ultralytics YOLO26. You can [train](../../modes/train.md) and [validate](../../modes/val.md), models directly using the provided YAML configuration file.
### Where can I find the kitti dataset configuration file?
You can access the YAML file at [https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/kitti.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/kitti.yaml).

159
algorithms/dms_yolo/code/docs/en/datasets/detect/lvis.md Normal file
View File

@@ -0,0 +1,159 @@
---
comments: true
description: Discover the LVIS dataset by Facebook AI Research, a benchmark for object detection and instance segmentation with a large, diverse vocabulary. Learn how to utilize it.
keywords: LVIS dataset, object detection, instance segmentation, Facebook AI Research, YOLO, computer vision, model training, LVIS examples
---
# LVIS Dataset
The [LVIS dataset](https://www.lvisdataset.org/) is a large-scale, fine-grained vocabulary-level annotation dataset developed and released by Facebook AI Research (FAIR). It is primarily used as a research benchmark for object detection and [instance segmentation](https://www.ultralytics.com/glossary/instance-segmentation) with a large vocabulary of categories, aiming to drive further advancements in computer vision field.
<p align="center">
<br>
<iframe loading="lazy" width="720" height="405" src="https://www.youtube.com/embed/cfTKj96TjSE"
title="YouTube video player" frameborder="0"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen>
</iframe>
<br>
<strong>Watch:</strong> YOLO World training workflow with LVIS dataset
</p>
<p align="center">
<img width="640" src="https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/lvis-dataset-example-images.avif" alt="LVIS large vocabulary instance segmentation dataset">
</p>
## Key Features
- LVIS contains 160k images and 2M instance annotations for object detection, segmentation, and captioning tasks.
- The dataset comprises 1203 object categories, including common objects like cars, bicycles, and animals, as well as more specific categories such as umbrellas, handbags, and sports equipment.
- Annotations include object bounding boxes, segmentation masks, and captions for each image.
- LVIS provides standardized evaluation metrics like [mean Average Precision](https://www.ultralytics.com/glossary/mean-average-precision-map) (mAP) for object detection, and mean Average [Recall](https://www.ultralytics.com/glossary/recall) (mAR) for segmentation tasks, making it suitable for comparing model performance.
- LVIS uses exactly the same images as [COCO](./coco.md) dataset, but with different splits and different annotations.
## Dataset Structure
The LVIS dataset is split into three subsets:
1. **Train**: This subset contains 100k images for training object detection, segmentation, and captioning models.
2. **Val**: This subset has 20k images used for validation purposes during model training.
3. **Minival**: This subset is exactly the same as COCO val2017 set which has 5k images used for validation purposes during model training.
4. **Test**: This subset consists of 20k images used for testing and benchmarking the trained models. Ground truth annotations for this subset are not publicly available, and the results are submitted to the [LVIS evaluation server](https://eval.ai/web/challenges/challenge-page/675/overview) for performance evaluation.
## Applications
The LVIS dataset is widely used for training and evaluating [deep learning](https://www.ultralytics.com/glossary/deep-learning-dl) models in object detection (such as [YOLO](../../models/yolo26.md), [Faster R-CNN](https://arxiv.org/abs/1506.01497), and [SSD](https://arxiv.org/abs/1512.02325)), instance segmentation (such as [Mask R-CNN](https://arxiv.org/abs/1703.06870)). The dataset's diverse set of object categories, large number of annotated images, and standardized evaluation metrics make it an essential resource for computer vision researchers and practitioners.
## Dataset YAML
A YAML (Yet Another Markup Language) file is used to define the dataset configuration. It contains information about the dataset's paths, classes, and other relevant information. In the case of the LVIS dataset, the `lvis.yaml` file is maintained at [https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/lvis.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/lvis.yaml).
!!! example "ultralytics/cfg/datasets/lvis.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/lvis.yaml"
```
## Usage
To train a YOLO26n model on the LVIS dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, you can use the following code snippets. For a comprehensive list of available arguments, refer to the model [Training](../../modes/train.md) page.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="lvis.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=lvis.yaml model=yolo26n.pt epochs=100 imgsz=640
```
## Sample Images and Annotations
The LVIS dataset contains a diverse set of images with various object categories and complex scenes. Here are some examples of images from the dataset, along with their corresponding annotations:
![LVIS large vocabulary instance segmentation dataset mosaic](https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/lvis-mosaiced-training-batch.avif)
- **Mosaiced Image**: This image demonstrates a training batch composed of mosaiced dataset images. Mosaicing is a technique used during training that combines multiple images into a single image to increase the variety of objects and scenes within each training batch. This helps improve the model's ability to generalize to different object sizes, aspect ratios, and contexts.
The example showcases the variety and complexity of the images in the LVIS dataset and the benefits of using mosaicing during the training process.
## Citations and Acknowledgments
If you use the LVIS dataset in your research or development work, please cite the following paper:
!!! quote ""
=== "BibTeX"
```bibtex
@inproceedings{gupta2019lvis,
title={LVIS: A Dataset for Large Vocabulary Instance Segmentation},
author={Gupta, Agrim and Dollar, Piotr and Girshick, Ross},
booktitle={Proceedings of the {IEEE} Conference on Computer Vision and Pattern Recognition},
year={2019}
}
```
We would like to acknowledge the LVIS Consortium for creating and maintaining this valuable resource for the [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv) community. For more information about the LVIS dataset and its creators, visit the [LVIS dataset website](https://www.lvisdataset.org/).
## FAQ
### What is the LVIS dataset, and how is it used in computer vision?
The [LVIS dataset](https://www.lvisdataset.org/) is a large-scale dataset with fine-grained vocabulary-level annotations developed by Facebook AI Research (FAIR). It is primarily used for object detection and instance segmentation, featuring over 1203 object categories and 2 million instance annotations. Researchers and practitioners use it to train and benchmark models like Ultralytics YOLO for advanced computer vision tasks. The dataset's extensive size and diversity make it an essential resource for pushing the boundaries of model performance in detection and segmentation.
### How can I train a YOLO26n model using the LVIS dataset?
To train a YOLO26n model on the LVIS dataset for 100 epochs with an image size of 640, follow the example below. This process utilizes Ultralytics' framework, which offers comprehensive training features.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="lvis.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=lvis.yaml model=yolo26n.pt epochs=100 imgsz=640
```
For detailed training configurations, refer to the [Training](../../modes/train.md) documentation.
### How does the LVIS dataset differ from the COCO dataset?
The images in the LVIS dataset are the same as those in the [COCO dataset](./coco.md), but the two differ in terms of splitting and annotations. LVIS provides a larger and more detailed vocabulary with 1203 object categories compared to COCO's 80 categories. Additionally, LVIS focuses on annotation completeness and diversity, aiming to push the limits of [object detection](https://www.ultralytics.com/glossary/object-detection) and instance segmentation models by offering more nuanced and comprehensive data.
### Why should I use Ultralytics YOLO for training on the LVIS dataset?
Ultralytics YOLO models, including the latest YOLO26, are optimized for real-time object detection with state-of-the-art [accuracy](https://www.ultralytics.com/glossary/accuracy) and speed. They support a wide range of annotations, such as the fine-grained ones provided by the LVIS dataset, making them ideal for advanced computer vision applications. Moreover, Ultralytics offers seamless integration with various [training](../../modes/train.md), [validation](../../modes/val.md), and [prediction](../../modes/predict.md) modes, ensuring efficient model development and deployment.
### Can I see some sample annotations from the LVIS dataset?
Yes, the LVIS dataset includes a variety of images with diverse object categories and complex scenes. Here is an example of a sample image along with its annotations:
![LVIS large vocabulary instance segmentation dataset mosaic](https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/lvis-mosaiced-training-batch.avif)
This mosaiced image demonstrates a training batch composed of multiple dataset images combined into one. Mosaicing increases the variety of objects and scenes within each training batch, enhancing the model's ability to generalize across different contexts. For more details on the LVIS dataset, explore the [LVIS dataset documentation](#key-features).

153
algorithms/dms_yolo/code/docs/en/datasets/detect/medical-pills.md Normal file
View File

@@ -0,0 +1,153 @@
---
comments: true
description: Explore the medical-pills detection dataset with labeled images. Essential for training AI models for pharmaceutical identification and automation.
keywords: medical-pills dataset, pill detection, pharmaceutical imaging, AI in healthcare, computer vision, object detection, medical automation, dataset for training
---
# Medical Pills Dataset
<a href="https://colab.research.google.com/github/ultralytics/notebooks/blob/main/notebooks/how-to-train-ultralytics-yolo-on-medical-pills-dataset.ipynb"><img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open Medical Pills Dataset In Colab"></a>
The medical-pills detection dataset is a proof-of-concept (POC) dataset, carefully curated to demonstrate the potential of AI in pharmaceutical applications. It contains labeled images specifically designed to train [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv) [models](https://docs.ultralytics.com/models/) for identifying medical-pills.
<p align="center">
<br>
<iframe loading="lazy" width="720" height="405" src="https://www.youtube.com/embed/8gePl_Zcs5c"
title="YouTube video player" frameborder="0"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen>
</iframe>
<br>
<strong>Watch:</strong> How to train Ultralytics YOLO26 Model on Medical Pills Detection Dataset in <a href="https://colab.research.google.com/github/ultralytics/notebooks/blob/main/notebooks/how-to-train-ultralytics-yolo-on-medical-pills-dataset.ipynb">Google Colab</a>
</p>
This dataset serves as a foundational resource for automating essential [tasks](https://docs.ultralytics.com/tasks/) such as quality control, packaging automation, and efficient sorting in pharmaceutical workflows. By integrating this dataset into projects, researchers and developers can explore innovative [solutions](https://docs.ultralytics.com/solutions/) that enhance [accuracy](https://www.ultralytics.com/glossary/accuracy), streamline operations, and ultimately contribute to improved healthcare outcomes.
## Dataset Structure
The medical-pills dataset is divided into two subsets:
- **Training set**: Consisting of 92 images, each annotated with the class `pill`.
- **Validation set**: Comprising 23 images with corresponding annotations.
## Applications
Using computer vision for medical-pills detection enables automation in the pharmaceutical industry, supporting tasks like:
- **Pharmaceutical Sorting**: Automating the sorting of pills based on size, shape, or color to enhance production efficiency.
- **AI Research and Development**: Serving as a benchmark for developing and testing computer vision algorithms in pharmaceutical use cases.
- **Digital Inventory Systems**: Powering smart inventory solutions by integrating automated pill recognition for real-time stock monitoring and replenishment planning.
- **Quality Control**: Ensuring consistency in pill production by identifying defects, irregularities, or contamination.
- **Counterfeit Detection**: Helping identify potentially counterfeit medications by analyzing visual characteristics against known standards.
## Dataset YAML
A YAML configuration file is provided to define the dataset's structure, including paths and classes. For the medical-pills dataset, the `medical-pills.yaml` file can be accessed at [https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/medical-pills.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/medical-pills.yaml).
!!! example "ultralytics/cfg/datasets/medical-pills.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/medical-pills.yaml"
```
## Usage
To train a YOLO26n model on the medical-pills dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, use the following examples. For detailed arguments, refer to the model's [Training](../../modes/train.md) page.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="medical-pills.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=medical-pills.yaml model=yolo26n.pt epochs=100 imgsz=640
```
!!! example "Inference Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("path/to/best.pt") # load a fine-tuned model
# Inference using the model
results = model.predict("https://ultralytics.com/assets/medical-pills-sample.jpg")
```
=== "CLI"
```bash
# Start prediction with a fine-tuned *.pt model
yolo detect predict model='path/to/best.pt' imgsz=640 source="https://ultralytics.com/assets/medical-pills-sample.jpg"
```
## Sample Images and Annotations
The medical-pills dataset features labeled images showcasing the diversity of pills. Below is an example of a labeled image from the dataset:
![Medical-pills dataset sample image](https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/medical-pills-dataset-sample-image.avif)
- **Mosaiced Image**: Displayed is a training batch comprising mosaiced dataset images. Mosaicing enhances training diversity by consolidating multiple images into one, improving model generalization.
## Integration with Other Datasets
For more comprehensive pharmaceutical analysis, consider combining the medical-pills dataset with other related datasets like [package-seg](../segment/package-seg.md) for packaging identification or medical imaging datasets like [brain-tumor](brain-tumor.md) to develop end-to-end healthcare AI solutions.
## Citations and Acknowledgments
The dataset is available under the [AGPL-3.0 License](https://github.com/ultralytics/ultralytics/blob/main/LICENSE).
If you use the Medical-pills dataset in your research or development work, please cite it using the mentioned details:
!!! quote ""
=== "BibTeX"
```bibtex
@dataset{Jocher_Ultralytics_Datasets_2024,
author = {Jocher, Glenn and Rizwan, Muhammad},
license = {AGPL-3.0},
month = {Dec},
title = {Ultralytics Datasets: Medical-pills Detection Dataset},
url = {https://docs.ultralytics.com/datasets/detect/medical-pills/},
version = {1.0.0},
year = {2024}
}
```
## FAQ
### What is the structure of the medical-pills dataset?
The dataset includes 92 images for training and 23 images for validation. Each image is annotated with the class `pill`, enabling effective training and evaluation of models for pharmaceutical applications.
### How can I train a YOLO26 model on the medical-pills dataset?
You can train a YOLO26 model for 100 epochs with an image size of 640px using the Python or CLI methods provided. Refer to the [Training Example](#usage) section for detailed instructions and check the [YOLO26 documentation](../../models/yolo26.md) for more information on model capabilities.
### What are the benefits of using the medical-pills dataset in AI projects?
The dataset enables automation in pill detection, contributing to counterfeit prevention, quality assurance, and pharmaceutical process optimization. It also serves as a valuable resource for developing AI solutions that can improve medication safety and supply chain efficiency.
### How do I perform inference on the medical-pills dataset?
Inference can be done using Python or CLI methods with a fine-tuned YOLO26 model. Refer to the [Inference Example](#usage) section for code snippets and the [Predict mode documentation](../../modes/predict.md) for additional options.
### Where can I find the YAML configuration file for the medical-pills dataset?
The YAML file is available at [medical-pills.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/medical-pills.yaml), containing dataset paths, classes, and additional configuration details essential for training models on this dataset.

152
algorithms/dms_yolo/code/docs/en/datasets/detect/objects365.md Normal file
View File

@@ -0,0 +1,152 @@
---
comments: true
description: Explore the Objects365 Dataset with 2M images and 30M bounding boxes across 365 categories. Enhance your object detection models with diverse, high-quality data.
keywords: Objects365 dataset, object detection, machine learning, deep learning, computer vision, annotated images, bounding boxes, YOLO26, high-resolution images, dataset configuration
---
# Objects365 Dataset
The [Objects365](https://www.objects365.org/) dataset is a large-scale, high-quality dataset designed to foster object detection research with a focus on diverse objects in the wild. Created by a team of [Megvii](https://en.megvii.com/) researchers, the dataset offers a wide range of high-resolution images with a comprehensive set of annotated bounding boxes covering 365 object categories.
<p align="center">
<br>
<iframe loading="lazy" width="720" height="405" src="https://www.youtube.com/embed/J-RH22rwx1A"
title="YouTube video player" frameborder="0"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen>
</iframe>
<br>
<strong>Watch:</strong> How to Train Ultralytics YOLO26 on the Objects365 Dataset with Ultralytics | 2M Annotations 🚀
</p>
## Key Features
- Objects365 contains 365 object categories, with 2 million images and over 30 million bounding boxes.
- The dataset includes diverse objects in various scenarios, providing a rich and challenging benchmark for object detection tasks.
- Annotations include bounding boxes for objects, making it suitable for training and evaluating object detection models.
- Objects365 pretrained models significantly outperform ImageNet pretrained models, leading to better generalization on various tasks.
## Dataset Structure
The Objects365 dataset is organized into a single set of images with corresponding annotations:
- **Images**: The dataset includes 2 million high-resolution images, each containing a variety of objects across 365 categories.
- **Annotations**: The images are annotated with over 30 million bounding boxes, providing comprehensive ground truth information for [object detection](https://docs.ultralytics.com/tasks/detect/) tasks.
## Applications
The Objects365 dataset is widely used for training and evaluating [deep learning](https://www.ultralytics.com/glossary/deep-learning-dl) models in object detection tasks. The dataset's diverse set of object categories and high-quality annotations make it a valuable resource for researchers and practitioners in the field of [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv).
## Dataset YAML
A YAML (Yet Another Markup Language) file is used to define the dataset configuration. It contains information about the dataset's paths, classes, and other relevant information. For the case of the Objects365 Dataset, the `Objects365.yaml` file is maintained at [https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/Objects365.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/Objects365.yaml).
!!! example "ultralytics/cfg/datasets/Objects365.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/Objects365.yaml"
```
## Usage
To train a YOLO26n model on the Objects365 dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, you can use the following code snippets. For a comprehensive list of available arguments, refer to the model [Training](../../modes/train.md) page.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="Objects365.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=Objects365.yaml model=yolo26n.pt epochs=100 imgsz=640
```
## Sample Data and Annotations
The Objects365 dataset contains a diverse set of high-resolution images with objects from 365 categories, providing rich context for [object detection](https://www.ultralytics.com/glossary/object-detection) tasks. Here are some examples of the images in the dataset:
![Objects365 dataset sample with diverse object annotations](https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/objects365-sample-image.avif)
- **Objects365**: This image demonstrates an example of object detection, where objects are annotated with bounding boxes. The dataset provides a wide range of images to facilitate the development of models for this task.
The example showcases the variety and complexity of the data in the Objects365 dataset and highlights the importance of accurate object detection for computer vision applications.
## Citations and Acknowledgments
If you use the Objects365 dataset in your research or development work, please cite the following paper:
!!! quote ""
=== "BibTeX"
```bibtex
@inproceedings{shao2019objects365,
title={Objects365: A Large-scale, High-quality Dataset for Object Detection},
author={Shao, Shuai and Li, Zeming and Zhang, Tianyuan and Peng, Chao and Yu, Gang and Li, Jing and Zhang, Xiangyu and Sun, Jian},
booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
pages={8425--8434},
year={2019}
}
```
We would like to acknowledge the team of researchers who created and maintain the Objects365 dataset as a valuable resource for the computer vision research community. For more information about the Objects365 dataset and its creators, visit the [Objects365 dataset website](https://www.objects365.org/).
## FAQ
### What is the Objects365 dataset used for?
The [Objects365 dataset](https://www.objects365.org/) is designed for object detection tasks in [machine learning](https://www.ultralytics.com/glossary/machine-learning-ml) and computer vision. It provides a large-scale, high-quality dataset with 2 million annotated images and 30 million bounding boxes across 365 categories. Leveraging such a diverse dataset helps improve the performance and generalization of object detection models, making it invaluable for research and development in the field.
### How can I train a YOLO26 model on the Objects365 dataset?
To train a YOLO26n model using the Objects365 dataset for 100 epochs with an image size of 640, follow these instructions:
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="Objects365.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=Objects365.yaml model=yolo26n.pt epochs=100 imgsz=640
```
Refer to the [Training](../../modes/train.md) page for a comprehensive list of available arguments.
### Why should I use the Objects365 dataset for my object detection projects?
The Objects365 dataset offers several advantages for object detection tasks:
1. **Diversity**: It includes 2 million images with objects in diverse scenarios, covering 365 categories.
2. **High-quality Annotations**: Over 30 million bounding boxes provide comprehensive ground truth data.
3. **Performance**: Models pretrained on Objects365 significantly outperform those trained on datasets like [ImageNet](https://docs.ultralytics.com/datasets/classify/imagenet/), leading to better generalization.
### Where can I find the YAML configuration file for the Objects365 dataset?
The YAML configuration file for the Objects365 dataset is available at [Objects365.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/Objects365.yaml). This file contains essential information such as dataset paths and class labels, crucial for setting up your training environment.
### How does the dataset structure of Objects365 enhance object detection modeling?
The [Objects365 dataset](https://www.objects365.org/) is organized with 2 million high-resolution images and comprehensive annotations of over 30 million bounding boxes. This structure ensures a robust dataset for training [deep learning](https://www.ultralytics.com/glossary/deep-learning-dl) models in object detection, offering a wide variety of objects and scenarios. Such diversity and volume help in developing models that are more accurate and capable of generalizing well to real-world applications. For more details on the dataset structure, refer to the [Dataset YAML](#dataset-yaml) section.

237
algorithms/dms_yolo/code/docs/en/datasets/detect/open-images-v7.md Normal file
View File

@@ -0,0 +1,237 @@
---
comments: true
description: Explore the comprehensive Open Images V7 dataset by Google. Learn about its annotations, applications, and use YOLO26 pretrained models for computer vision tasks.
keywords: Open Images V7, Google dataset, computer vision, YOLO26 models, object detection, image segmentation, visual relationships, AI research, Ultralytics
---
# Open Images V7 Dataset
[Open Images V7](https://storage.googleapis.com/openimages/web/index.html) is a versatile and expansive dataset championed by Google. Aimed at propelling research in the realm of [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv), it boasts a vast collection of images annotated with a plethora of data, including image-level labels, object bounding boxes, object segmentation masks, visual relationships, and localized narratives.
<p align="center">
<br>
<iframe loading="lazy" width="720" height="405" src="https://www.youtube.com/embed/u3pLlgzUeV8"
title="YouTube video player" frameborder="0"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen>
</iframe>
<br>
<strong>Watch:</strong> <a href="https://www.ultralytics.com/glossary/object-detection">Object Detection</a> using OpenImagesV7 Pretrained Model
</p>
## Open Images V7 Pretrained Models
| Model | size<br><sup>(pixels)</sup> | mAP<sup>val<br>50-95</sup> | Speed<br><sup>CPU ONNX<br>(ms)</sup> | Speed<br><sup>A100 TensorRT<br>(ms)</sup> | params<br><sup>(M)</sup> | FLOPs<br><sup>(B)</sup> |
| ----------------------------------------------------------------------------------------- | --------------------------- | -------------------------- | ------------------------------------ | ----------------------------------------- | ------------------------ | ----------------------- |
| [YOLOv8n](https://github.com/ultralytics/assets/releases/download/v8.4.0/yolov8n-oiv7.pt) | 640 | 18.4 | 142.4 | 1.21 | 3.5 | 10.5 |
| [YOLOv8s](https://github.com/ultralytics/assets/releases/download/v8.4.0/yolov8s-oiv7.pt) | 640 | 27.7 | 183.1 | 1.40 | 11.4 | 29.7 |
| [YOLOv8m](https://github.com/ultralytics/assets/releases/download/v8.4.0/yolov8m-oiv7.pt) | 640 | 33.6 | 408.5 | 2.26 | 26.2 | 80.6 |
| [YOLOv8l](https://github.com/ultralytics/assets/releases/download/v8.4.0/yolov8l-oiv7.pt) | 640 | 34.9 | 596.9 | 2.43 | 44.1 | 167.4 |
| [YOLOv8x](https://github.com/ultralytics/assets/releases/download/v8.4.0/yolov8x-oiv7.pt) | 640 | 36.3 | 860.6 | 3.56 | 68.7 | 260.6 |
You can use these pretrained models for inference or fine-tuning as follows.
!!! example "Pretrained Model Usage Example"
=== "Python"
```python
from ultralytics import YOLO
# Load an Open Images Dataset V7 pretrained YOLOv8n model
model = YOLO("yolov8n-oiv7.pt")
# Run prediction
results = model.predict(source="image.jpg")
# Start training from the pretrained checkpoint
results = model.train(data="coco8.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Predict using an Open Images Dataset V7 pretrained model
yolo detect predict source=image.jpg model=yolov8n-oiv7.pt
# Start training from an Open Images Dataset V7 pretrained checkpoint
yolo detect train data=coco8.yaml model=yolov8n-oiv7.pt epochs=100 imgsz=640
```
![Open Images V7 classes visual](https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/open-images-v7-classes-visual.avif)
## Key Features
- Encompasses ~9M images annotated in various ways to suit multiple computer vision tasks.
- Houses a staggering 16M bounding boxes across 600 object classes in 1.9M images. These boxes are primarily hand-drawn by experts ensuring high [precision](https://www.ultralytics.com/glossary/precision).
- Visual relationship annotations totaling 3.3M are available, detailing 1,466 unique relationship triplets, object properties, and human activities.
- V5 introduced segmentation masks for 2.8M objects across 350 classes.
- V6 introduced 675k localized narratives that amalgamate voice, text, and mouse traces highlighting described objects.
- V7 introduced 66.4M point-level labels on 1.4M images, spanning 5,827 classes.
- Encompasses 61.4M image-level labels across a diverse set of 20,638 classes.
- Provides a unified platform for [image classification](https://www.ultralytics.com/glossary/image-classification), object detection, relationship detection, [instance segmentation](https://www.ultralytics.com/glossary/instance-segmentation), and multimodal image descriptions.
## Dataset Structure
Open Images V7 is structured in multiple components catering to varied computer vision challenges:
- **Images**: About 9 million images, often showcasing intricate scenes with an average of 8.3 objects per image.
- **Bounding Boxes**: Over 16 million boxes that demarcate objects across 600 categories.
- **Segmentation Masks**: These detail the exact boundary of 2.8M objects across 350 classes.
- **Visual Relationships**: 3.3M annotations indicating object relationships, properties, and actions.
- **Localized Narratives**: 675k descriptions combining voice, text, and mouse traces.
- **Point-Level Labels**: 66.4M labels across 1.4M images, suitable for zero/few-shot [semantic segmentation](https://www.ultralytics.com/glossary/semantic-segmentation).
## Applications
Open Images V7 is a cornerstone for training and evaluating state-of-the-art models in various computer vision tasks. The dataset's broad scope and high-quality annotations make it indispensable for researchers and developers specializing in [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv).
Some key applications include:
- **Advanced Object Detection**: Train models to identify and locate multiple objects in complex scenes with high accuracy.
- **Semantic Understanding**: Develop systems that comprehend visual relationships between objects.
- **Image Segmentation**: Create precise pixel-level masks for objects, enabling detailed scene analysis.
- **Multi-modal Learning**: Combine visual data with text descriptions for richer AI understanding.
- **Zero-shot Learning**: Leverage the extensive class coverage to identify objects not seen during training.
## Dataset YAML
Ultralytics maintains an `open-images-v7.yaml` file that specifies the dataset paths, class names, and other configuration details required for training.
!!! example "OpenImagesV7.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/open-images-v7.yaml"
```
## Usage
To train a YOLO26n model on the Open Images V7 dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, you can use the following code snippets. For a comprehensive list of available arguments, refer to the model [Training](../../modes/train.md) page.
!!! warning
The complete Open Images V7 dataset comprises 1,743,042 training images and 41,620 validation images, requiring approximately **561 GB of storage space** upon download.
Executing the commands provided below will trigger an automatic download of the full dataset if it's not already present locally. Before running the below example it's crucial to:
- Verify that your device has enough storage capacity.
- Ensure a robust and speedy internet connection.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a COCO-pretrained YOLO26n model
model = YOLO("yolo26n.pt")
# Train the model on the Open Images V7 dataset
results = model.train(data="open-images-v7.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Train a COCO-pretrained YOLO26n model on the Open Images V7 dataset
yolo detect train data=open-images-v7.yaml model=yolo26n.pt epochs=100 imgsz=640
```
## Sample Data and Annotations
Illustrations of the dataset help provide insights into its richness:
![Open Images V7 dataset sample with bounding box annotations](https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/oidv7-all-in-one-example-ab.avif)
- **Open Images V7**: This image exemplifies the depth and detail of annotations available, including bounding boxes, relationships, and segmentation masks.
Researchers can gain invaluable insights into the array of computer vision challenges that the dataset addresses, from basic object detection to intricate relationship identification. The [diversity of annotations](https://docs.ultralytics.com/datasets/explorer/) makes Open Images V7 particularly valuable for developing models that can understand complex visual scenes.
## Citations and Acknowledgments
For those employing Open Images V7 in their work, it's prudent to cite the relevant papers and acknowledge the creators:
!!! quote ""
=== "BibTeX"
```bibtex
@article{OpenImages,
author = {Alina Kuznetsova and Hassan Rom and Neil Alldrin and Jasper Uijlings and Ivan Krasin and Jordi Pont-Tuset and Shahab Kamali and Stefan Popov and Matteo Malloci and Alexander Kolesnikov and Tom Duerig and Vittorio Ferrari},
title = {The Open Images Dataset V4: Unified image classification, object detection, and visual relationship detection at scale},
year = {2020},
journal = {IJCV}
}
```
A heartfelt acknowledgment goes out to the Google AI team for creating and maintaining the Open Images V7 dataset. For a deep dive into the dataset and its offerings, navigate to the [official Open Images V7 website](https://storage.googleapis.com/openimages/web/index.html).
## FAQ
### What is the Open Images V7 dataset?
Open Images V7 is an extensive and versatile dataset created by Google, designed to advance research in computer vision. It includes image-level labels, object bounding boxes, object segmentation masks, visual relationships, and localized narratives, making it ideal for various computer vision tasks such as object detection, segmentation, and relationship detection.
### How do I train a YOLO26 model on the Open Images V7 dataset?
To train a YOLO26 model on the Open Images V7 dataset, you can use both Python and CLI commands. Here's an example of training the YOLO26n model for 100 epochs with an image size of 640:
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a COCO-pretrained YOLO26n model
model = YOLO("yolo26n.pt")
# Train the model on the Open Images V7 dataset
results = model.train(data="open-images-v7.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Train a COCO-pretrained YOLO26n model on the Open Images V7 dataset
yolo detect train data=open-images-v7.yaml model=yolo26n.pt epochs=100 imgsz=640
```
For more details on arguments and settings, refer to the [Training](../../modes/train.md) page.
### What are some key features of the Open Images V7 dataset?
The Open Images V7 dataset includes approximately 9 million images with various annotations:
- **Bounding Boxes**: 16 million bounding boxes across 600 object classes.
- **Segmentation Masks**: Masks for 2.8 million objects across 350 classes.
- **Visual Relationships**: 3.3 million annotations indicating relationships, properties, and actions.
- **Localized Narratives**: 675,000 descriptions combining voice, text, and mouse traces.
- **Point-Level Labels**: 66.4 million labels across 1.4 million images.
- **Image-Level Labels**: 61.4 million labels across 20,638 classes.
### What pretrained models are available for the Open Images V7 dataset?
Ultralytics provides several YOLOv8 pretrained models for the Open Images V7 dataset, each with different sizes and performance metrics:
| Model | size<br><sup>(pixels)</sup> | mAP<sup>val<br>50-95</sup> | Speed<br><sup>CPU ONNX<br>(ms)</sup> | Speed<br><sup>A100 TensorRT<br>(ms)</sup> | params<br><sup>(M)</sup> | FLOPs<br><sup>(B)</sup> |
| ----------------------------------------------------------------------------------------- | --------------------------- | -------------------------- | ------------------------------------ | ----------------------------------------- | ------------------------ | ----------------------- |
| [YOLOv8n](https://github.com/ultralytics/assets/releases/download/v8.4.0/yolov8n-oiv7.pt) | 640 | 18.4 | 142.4 | 1.21 | 3.5 | 10.5 |
| [YOLOv8s](https://github.com/ultralytics/assets/releases/download/v8.4.0/yolov8s-oiv7.pt) | 640 | 27.7 | 183.1 | 1.40 | 11.4 | 29.7 |
| [YOLOv8m](https://github.com/ultralytics/assets/releases/download/v8.4.0/yolov8m-oiv7.pt) | 640 | 33.6 | 408.5 | 2.26 | 26.2 | 80.6 |
| [YOLOv8l](https://github.com/ultralytics/assets/releases/download/v8.4.0/yolov8l-oiv7.pt) | 640 | 34.9 | 596.9 | 2.43 | 44.1 | 167.4 |
| [YOLOv8x](https://github.com/ultralytics/assets/releases/download/v8.4.0/yolov8x-oiv7.pt) | 640 | 36.3 | 860.6 | 3.56 | 68.7 | 260.6 |
### What applications can the Open Images V7 dataset be used for?
The Open Images V7 dataset supports a variety of computer vision tasks including:
- **[Image Classification](https://www.ultralytics.com/glossary/image-classification)**
- **Object Detection**
- **Instance Segmentation**
- **Visual Relationship Detection**
- **Multimodal Image Descriptions**
Its comprehensive annotations and broad scope make it suitable for training and evaluating advanced [machine learning](https://www.ultralytics.com/glossary/machine-learning-ml) models, as highlighted in practical use cases detailed in our [applications](#applications) section.

182
algorithms/dms_yolo/code/docs/en/datasets/detect/roboflow-100.md Normal file
View File

@@ -0,0 +1,182 @@
---
comments: true
description: Explore the Roboflow 100 dataset featuring 100 diverse datasets designed to test object detection models across various domains, from healthcare to video games.
keywords: Roboflow 100, Ultralytics, object detection, dataset, benchmarking, machine learning, computer vision, diverse datasets, model evaluation
---
# Roboflow 100 Dataset
Roboflow 100, sponsored by [Intel](https://www.intel.com/), is a groundbreaking [object detection](../../tasks/detect.md) benchmark dataset. It includes 100 diverse datasets sampled from over 90,000 public datasets available on Roboflow Universe. This benchmark is specifically designed to test the adaptability of [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv) models, like [Ultralytics YOLO models](../../models/yolo26.md), to various domains, including healthcare, aerial imagery, and video games.
!!! question "Licensing"
Ultralytics offers two licensing options to accommodate different use cases:
- **AGPL-3.0 License**: This [OSI-approved](https://opensource.org/license) open-source license is ideal for students and enthusiasts, promoting open collaboration and knowledge sharing. See the [LICENSE](https://github.com/ultralytics/ultralytics/blob/main/LICENSE) file for more details and visit our [AGPL-3.0 License page](https://www.ultralytics.com/legal/agpl-3-0-software-license).
- **Enterprise License**: Designed for commercial use, this license allows for the seamless integration of Ultralytics software and AI models into commercial products and services. If your scenario involves commercial applications, please reach out via [Ultralytics Licensing](https://www.ultralytics.com/license).
<p align="center">
<img width="640" src="https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/roboflow-100-overview.avif" alt="Roboflow 100 diverse object detection benchmark">
</p>
## Key Features
- **Diverse Domains**: Includes 100 datasets across seven distinct domains: Aerial, Video games, Microscopic, Underwater, Documents, Electromagnetic, and Real World.
- **Scale**: The benchmark comprises 224,714 images across 805 classes, representing over 11,170 hours of [data labeling](https://www.ultralytics.com/glossary/data-labeling) effort.
- **Standardization**: All images are [preprocessed](https://www.ultralytics.com/glossary/data-preprocessing) and resized to 640x640 pixels for consistent evaluation.
- **Clean Evaluation**: Focuses on eliminating class ambiguity and filters out underrepresented classes to ensure cleaner [model evaluation](../../guides/model-evaluation-insights.md).
- **Annotations**: Includes [bounding boxes](https://www.ultralytics.com/glossary/bounding-box) for objects, suitable for [training](../../modes/train.md) and evaluating object detection models using metrics like [mAP](https://www.ultralytics.com/glossary/mean-average-precision-map).
## Dataset Structure
The Roboflow 100 dataset is organized into seven categories, each containing a unique collection of datasets, images, and classes:
- **Aerial**: 7 datasets, 9,683 images, 24 classes.
- **Video Games**: 7 datasets, 11,579 images, 88 classes.
- **Microscopic**: 11 datasets, 13,378 images, 28 classes.
- **Underwater**: 5 datasets, 18,003 images, 39 classes.
- **Documents**: 8 datasets, 24,813 images, 90 classes.
- **Electromagnetic**: 12 datasets, 36,381 images, 41 classes.
- **Real World**: 50 datasets, 110,615 images, 495 classes.
This structure provides a diverse and extensive testing ground for [object detection](https://www.ultralytics.com/glossary/object-detection) models, reflecting a wide array of real-world application scenarios found in various [Ultralytics Solutions](https://www.ultralytics.com/solutions).
## Benchmarking
Dataset [benchmarking](../../modes/benchmark.md) involves evaluating the performance of [machine learning](https://www.ultralytics.com/glossary/machine-learning-ml) models on specific datasets using standardized metrics. Common metrics include [accuracy](https://www.ultralytics.com/glossary/accuracy), mean Average Precision (mAP), and [F1-score](https://www.ultralytics.com/glossary/f1-score). You can learn more about these in our [YOLO Performance Metrics guide](../../guides/yolo-performance-metrics.md).
!!! tip "Benchmarking Results"
Benchmarking results using the provided script will be stored in the `ultralytics-benchmarks/` directory, specifically in `evaluation.txt`.
!!! example "Benchmarking Example"
The following script demonstrates how to programmatically benchmark an Ultralytics YOLO model (e.g., YOLO26n) on all 100 datasets within the Roboflow 100 benchmark using the `RF100Benchmark` class.
=== "Python"
```python
import os
import shutil
from pathlib import Path
from ultralytics.utils.benchmarks import RF100Benchmark
# Initialize RF100Benchmark and set API key
benchmark = RF100Benchmark()
benchmark.set_key(api_key="YOUR_ROBOFLOW_API_KEY")
# Parse dataset and define file paths
names, cfg_yamls = benchmark.parse_dataset()
val_log_file = Path("ultralytics-benchmarks") / "validation.txt"
eval_log_file = Path("ultralytics-benchmarks") / "evaluation.txt"
# Run benchmarks on each dataset in RF100
for ind, path in enumerate(cfg_yamls):
path = Path(path)
if path.exists():
# Fix YAML file and run training
benchmark.fix_yaml(str(path))
os.system(f"yolo detect train data={path} model=yolo26s.pt epochs=1 batch=16")
# Run validation and evaluate
os.system(f"yolo detect val data={path} model=runs/detect/train/weights/best.pt > {val_log_file} 2>&1")
benchmark.evaluate(str(path), str(val_log_file), str(eval_log_file), ind)
# Remove the 'runs' directory
runs_dir = Path.cwd() / "runs"
shutil.rmtree(runs_dir)
else:
print("YAML file path does not exist")
continue
print("RF100 Benchmarking completed!")
```
## Applications
Roboflow 100 is invaluable for various applications related to [computer vision](https://www.ultralytics.com/blog/everything-you-need-to-know-about-computer-vision-in-2025) and [deep learning](https://www.ultralytics.com/glossary/deep-learning-dl). Researchers and engineers can leverage this benchmark to:
- Evaluate the performance of object detection models in a multi-domain context.
- Test the adaptability and [robustness](<https://en.wikipedia.org/wiki/Robustness_(computer_science)>) of models to real-world scenarios beyond common [benchmark datasets](https://www.ultralytics.com/glossary/benchmark-dataset) like [COCO](https://cocodataset.org/#home) or [PASCAL VOC](http://host.robots.ox.ac.uk/pascal/VOC/).
- Benchmark the capabilities of object detection models across diverse datasets, including specialized areas like healthcare, aerial imagery, and video games.
- Compare model performance across different [neural network](https://www.ultralytics.com/glossary/neural-network-nn) architectures and [optimization](https://www.ultralytics.com/glossary/optimization-algorithm) techniques.
- Identify domain-specific challenges that may require specialized [model training tips](../../guides/model-training-tips.md) or [fine-tuning](https://www.ultralytics.com/glossary/fine-tuning) approaches like [transfer learning](https://www.ultralytics.com/glossary/transfer-learning).
For more ideas and inspiration on real-world applications, explore [our guides on practical projects](../../guides/index.md) or check out [Ultralytics Platform](https://platform.ultralytics.com) for streamlined [model training](../../modes/train.md) and [deployment](../../guides/model-deployment-options.md).
## Usage
The Roboflow 100 dataset, including metadata and download links, is available on the official [Roboflow 100 GitHub repository](https://github.com/roboflow/roboflow-100-benchmark). You can access and utilize the dataset directly from there for your benchmarking needs. The Ultralytics `RF100Benchmark` utility simplifies the process of downloading and preparing these datasets for use with Ultralytics models.
## Sample Data and Annotations
Roboflow 100 consists of datasets with diverse images captured from various angles and domains. Below are examples of annotated images included in the RF100 benchmark, showcasing the variety of objects and scenes. Techniques like [data augmentation](https://www.ultralytics.com/glossary/data-augmentation) can further enhance the diversity during training.
<p align="center">
<img width="640" src="https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/sample-data-annotations.avif" alt="Roboflow 100 sample images with annotations">
</p>
The diversity seen in the Roboflow 100 benchmark represents a significant advancement from traditional benchmarks, which often focus on optimizing a single metric within a limited domain. This comprehensive approach aids in developing more robust and versatile [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv) models capable of performing well across a multitude of different scenarios.
## Citations and Acknowledgments
If you use the Roboflow 100 dataset in your research or development work, please cite the original paper:
!!! quote ""
=== "BibTeX"
```bibtex
@misc{rf100benchmark,
Author = {Floriana Ciaglia and Francesco Saverio Zuppichini and Paul Guerrie and Mark McQuade and Jacob Solawetz},
Title = {Roboflow 100: A Rich, Multi-Domain Object Detection Benchmark},
Year = {2022},
Eprint = {arXiv:2211.13523},
url = {https://arxiv.org/abs/2211.13523}
}
```
We extend our gratitude to the Roboflow team and all contributors for their significant efforts in creating and maintaining the Roboflow 100 dataset as a valuable resource for the computer vision community.
If you are interested in exploring more datasets to enhance your object detection and machine learning projects, feel free to visit [our comprehensive dataset collection](../index.md), which includes a variety of other [detection datasets](../detect/index.md).
## FAQ
### What is the Roboflow 100 dataset, and why is it significant for object detection?
The **Roboflow 100** dataset is a benchmark for [object detection](../../tasks/detect.md) models. It comprises 100 diverse datasets sourced from Roboflow Universe, covering domains like healthcare, aerial imagery, and video games. Its significance lies in providing a standardized way to test model adaptability and robustness across a wide range of real-world scenarios, moving beyond traditional, often domain-limited, benchmarks.
### Which domains are covered by the Roboflow 100 dataset?
The **Roboflow 100** dataset spans seven diverse domains, offering unique challenges for [object detection](https://www.ultralytics.com/glossary/object-detection) models:
1. **Aerial**: 7 datasets (e.g., satellite imagery, drone views).
2. **Video Games**: 7 datasets (e.g., objects from various game environments).
3. **Microscopic**: 11 datasets (e.g., cells, particles).
4. **Underwater**: 5 datasets (e.g., marine life, submerged objects).
5. **Documents**: 8 datasets (e.g., text regions, form elements).
6. **Electromagnetic**: 12 datasets (e.g., radar signatures, spectral data visualizations).
7. **Real World**: 50 datasets (a broad category including everyday objects, scenes, retail, etc.).
This variety makes RF100 an excellent resource for assessing the [generalizability](<https://en.wikipedia.org/wiki/Generalization_(learning)>) of computer vision models.
### What should I include when citing the Roboflow 100 dataset in my research?
When using the Roboflow 100 dataset, please cite the original paper to give credit to the creators. Here is the recommended BibTeX citation:
!!! quote ""
=== "BibTeX"
```bibtex
@misc{rf100benchmark,
Author = {Floriana Ciaglia and Francesco Saverio Zuppichini and Paul Guerrie and Mark McQuade and Jacob Solawetz},
Title = {Roboflow 100: A Rich, Multi-Domain Object Detection Benchmark},
Year = {2022},
Eprint = {arXiv:2211.13523},
url = {https://arxiv.org/abs/2211.13523}
}
```
For further exploration, consider visiting our [comprehensive dataset collection](../index.md) or browsing other [detection datasets](../detect/index.md) compatible with Ultralytics models.

179
algorithms/dms_yolo/code/docs/en/datasets/detect/signature.md Normal file
View File

@@ -0,0 +1,179 @@
---
comments: true
description: Discover the Signature Detection Dataset for training models to identify and verify human signatures in various documents. Perfect for document verification and fraud prevention.
keywords: Signature Detection Dataset, document verification, fraud detection, computer vision, YOLO26, Ultralytics, annotated signatures, training dataset
---
# Signature Detection Dataset
This dataset focuses on detecting human written signatures within documents. It includes a variety of document types with annotated signatures, providing valuable insights for applications in document verification and fraud detection. Essential for training [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv) algorithms, this dataset aids in identifying signatures in various document formats, supporting research and practical applications in document analysis.
## Dataset Structure
The signature detection dataset is split into two subsets:
- **Training set**: Contains 143 images, each with corresponding annotations.
- **Validation set**: Includes 35 images, each with paired annotations.
## Applications
This dataset can be applied in various computer vision tasks such as [object detection](https://www.ultralytics.com/glossary/object-detection), [object tracking](https://docs.ultralytics.com/modes/track/), and document analysis. Specifically, it can be used to train and evaluate models for identifying signatures in documents, which has significant applications in:
- **Document Verification**: Automating the verification process for legal and financial documents
- **Fraud Detection**: Identifying potentially forged or unauthorized signatures
- **Digital Document Processing**: Streamlining workflows in administrative and legal sectors
- **Banking and Finance**: Enhancing security in check processing and loan document verification
- **Archival Research**: Supporting historical document analysis and cataloging
Additionally, it serves as a valuable resource for educational purposes, enabling students and researchers to study signature characteristics across different document types.
## Dataset YAML
A YAML (Yet Another Markup Language) file defines the dataset configuration, including paths and classes information. For the signature detection dataset, the `signature.yaml` file is located at [https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/signature.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/signature.yaml).
!!! example "ultralytics/cfg/datasets/signature.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/signature.yaml"
```
## Usage
To train a YOLO26n model on the signature detection dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, use the provided code samples. For a comprehensive list of available parameters, refer to the model's [Training](../../modes/train.md) page.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="signature.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=signature.yaml model=yolo26n.pt epochs=100 imgsz=640
```
!!! example "Inference Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("path/to/best.pt") # load a signature-detection fine-tuned model
# Inference using the model
results = model.predict("https://ultralytics.com/assets/signature-s.mp4", conf=0.75)
```
=== "CLI"
```bash
# Start prediction with a finetuned *.pt model
yolo detect predict model='path/to/best.pt' imgsz=640 source="https://ultralytics.com/assets/signature-s.mp4" conf=0.75
```
## Sample Images and Annotations
The signature detection dataset comprises a wide variety of images showcasing different document types and annotated signatures. Below are examples of images from the dataset, each accompanied by its corresponding annotations.
![Signature detection dataset sample image](https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/signature-detection-mosaiced-sample.avif)
- **Mosaiced Image**: Here, we present a training batch consisting of mosaiced dataset images. Mosaicing, a training technique, combines multiple images into one, enriching batch diversity. This method helps enhance the model's ability to generalize across different signature sizes, aspect ratios, and contexts.
This example illustrates the variety and complexity of images in the signature Detection Dataset, emphasizing the benefits of including mosaicing during the training process.
## Citations and Acknowledgments
The dataset has been released available under the [AGPL-3.0 License](https://github.com/ultralytics/ultralytics/blob/main/LICENSE).
## FAQ
### What is the Signature Detection Dataset, and how can it be used?
The Signature Detection Dataset is a collection of annotated images aimed at detecting human signatures within various document types. It can be applied in computer vision tasks such as [object detection](https://www.ultralytics.com/glossary/object-detection) and tracking, primarily for document verification, fraud detection, and archival research. This dataset helps train models to recognize signatures in different contexts, making it valuable for both research and practical applications in [smart document analysis](https://www.ultralytics.com/blog/using-ultralytics-yolo11-for-smart-document-analysis).
### How do I train a YOLO26n model on the Signature Detection Dataset?
To train a YOLO26n model on the Signature Detection Dataset, follow these steps:
1. Download the `signature.yaml` dataset configuration file from [signature.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/signature.yaml).
2. Use the following Python script or CLI command to start training:
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a pretrained model
model = YOLO("yolo26n.pt")
# Train the model
results = model.train(data="signature.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
yolo detect train data=signature.yaml model=yolo26n.pt epochs=100 imgsz=640
```
For more details, refer to the [Training](../../modes/train.md) page.
### What are the main applications of the Signature Detection Dataset?
The Signature Detection Dataset can be used for:
1. **Document Verification**: Automatically verifying the presence and authenticity of human signatures in documents.
2. **Fraud Detection**: Identifying forged or fraudulent signatures in legal and financial documents.
3. **Archival Research**: Assisting historians and archivists in the digital analysis and cataloging of historical documents.
4. **Education**: Supporting academic research and teaching in the fields of computer vision and [machine learning](https://www.ultralytics.com/glossary/machine-learning-ml).
5. **Financial Services**: Enhancing security in banking transactions and loan processing by verifying signature authenticity.
### How can I perform inference using a model trained on the Signature Detection Dataset?
To perform inference using a model trained on the Signature Detection Dataset, follow these steps:
1. Load your fine-tuned model.
2. Use the below Python script or CLI command to perform inference:
!!! example "Inference Example"
=== "Python"
```python
from ultralytics import YOLO
# Load the fine-tuned model
model = YOLO("path/to/best.pt")
# Perform inference
results = model.predict("https://ultralytics.com/assets/signature-s.mp4", conf=0.75)
```
=== "CLI"
```bash
yolo detect predict model='path/to/best.pt' imgsz=640 source="https://ultralytics.com/assets/signature-s.mp4" conf=0.75
```
### What is the structure of the Signature Detection Dataset, and where can I find more information?
The Signature Detection Dataset is divided into two subsets:
- **Training Set**: Contains 143 images with annotations.
- **Validation Set**: Includes 35 images with annotations.
For detailed information, you can refer to the [Dataset Structure](#dataset-structure) section. Additionally, view the complete dataset configuration in the `signature.yaml` file located at [signature.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/signature.yaml).

189
algorithms/dms_yolo/code/docs/en/datasets/detect/sku-110k.md Normal file
View File

@@ -0,0 +1,189 @@
---
comments: true
description: Explore the SKU-110k dataset of densely packed retail shelf images, perfect for training and evaluating deep learning models in object detection tasks.
keywords: SKU-110k, dataset, object detection, retail shelf images, deep learning, computer vision, model training
---
# SKU-110k Dataset
The [SKU-110k](https://github.com/eg4000/SKU110K_CVPR19) dataset is a collection of densely packed retail shelf images, designed to support research in [object detection](https://www.ultralytics.com/glossary/object-detection) tasks. Developed by Eran Goldman et al., the dataset contains over 110,000 unique store keeping unit (SKU) categories with densely packed objects, often looking similar or even identical, positioned in proximity.
<p align="center">
<br>
<iframe loading="lazy" width="720" height="405" src="https://www.youtube.com/embed/_gRqR-miFPE"
title="YouTube video player" frameborder="0"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen>
</iframe>
<br>
<strong>Watch:</strong> How to Train YOLOv10 on SKU-110k Dataset using Ultralytics | Retail Dataset
</p>
![SKU-110K dataset densely packed retail shelf detection](https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/densely-packed-retail-shelf.avif)
## Key Features
- SKU-110k contains images of store shelves from around the world, featuring densely packed objects that pose challenges for state-of-the-art object detectors.
- The dataset includes over 110,000 unique SKU categories, providing a diverse range of object appearances.
- Annotations include bounding boxes for objects and SKU category labels.
## Dataset Structure
The SKU-110k dataset is organized into three main subsets:
1. **Training set**: This subset contains 8,219 images and annotations used for training object detection models.
2. **Validation set**: This subset consists of 588 images and annotations used for model validation during training.
3. **Test set**: This subset includes 2,936 images designed for the final evaluation of trained object detection models.
## Applications
The SKU-110k dataset is widely used for training and evaluating [deep learning](https://www.ultralytics.com/glossary/deep-learning-dl) models in object detection tasks, especially in densely packed scenes such as retail shelf displays. Its applications include:
- Retail inventory management and automation
- Product recognition in e-commerce platforms
- Planogram compliance verification
- Self-checkout systems in stores
- Robotic picking and sorting in warehouses
The dataset's diverse set of SKU categories and densely packed object arrangements make it a valuable resource for researchers and practitioners in the field of [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv).
## Dataset YAML
A YAML (Yet Another Markup Language) file is used to define the dataset configuration. It contains information about the dataset's paths, classes, and other relevant information. For the case of the SKU-110K dataset, the `SKU-110K.yaml` file is maintained at [https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/SKU-110K.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/SKU-110K.yaml).
!!! example "ultralytics/cfg/datasets/SKU-110K.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/SKU-110K.yaml"
```
## Usage
To train a YOLO26n model on the SKU-110K dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, you can use the following code snippets. For a comprehensive list of available arguments, refer to the model [Training](../../modes/train.md) page.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="SKU-110K.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=SKU-110K.yaml model=yolo26n.pt epochs=100 imgsz=640
```
## Sample Data and Annotations
The SKU-110k dataset contains a diverse set of retail shelf images with densely packed objects, providing rich context for object detection tasks. Here are some examples of data from the dataset, along with their corresponding annotations:
![SKU-110K retail product detection on store shelves](https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/densely-packed-retail-shelf-1.avif)
- **Densely packed retail shelf image**: This image demonstrates an example of densely packed objects in a retail shelf setting. Objects are annotated with bounding boxes and SKU category labels.
The example showcases the variety and complexity of the data in the SKU-110k dataset and highlights the importance of high-quality data for object detection tasks. The dense arrangement of products presents unique challenges for detection algorithms, making this dataset particularly valuable for developing robust retail-focused computer vision solutions.
## Citations and Acknowledgments
If you use the SKU-110k dataset in your research or development work, please cite the following paper:
!!! quote ""
=== "BibTeX"
```bibtex
@inproceedings{goldman2019dense,
author = {Eran Goldman and Roei Herzig and Aviv Eisenschtat and Jacob Goldberger and Tal Hassner},
title = {Precise Detection in Densely Packed Scenes},
booktitle = {Proc. Conf. Comput. Vision Pattern Recognition (CVPR)},
year = {2019}
}
```
We would like to acknowledge Eran Goldman et al. for creating and maintaining the SKU-110k dataset as a valuable resource for the computer vision research community. For more information about the SKU-110k dataset and its creators, visit the [SKU-110k dataset GitHub repository](https://github.com/eg4000/SKU110K_CVPR19).
## FAQ
### What is the SKU-110k dataset and why is it important for object detection?
The SKU-110k dataset consists of densely packed retail shelf images designed to aid research in object detection tasks. Developed by Eran Goldman et al., it includes over 110,000 unique SKU categories. Its importance lies in its ability to challenge state-of-the-art object detectors with diverse object appearances and proximity, making it an invaluable resource for researchers and practitioners in computer vision. Learn more about the dataset's structure and applications in our [SKU-110k Dataset](#sku-110k-dataset) section.
### How do I train a YOLO26 model using the SKU-110k dataset?
Training a YOLO26 model on the SKU-110k dataset is straightforward. Here's an example to train a YOLO26n model for 100 epochs with an image size of 640:
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="SKU-110K.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=SKU-110K.yaml model=yolo26n.pt epochs=100 imgsz=640
```
For a comprehensive list of available arguments, refer to the model [Training](../../modes/train.md) page.
### What are the main subsets of the SKU-110k dataset?
The SKU-110k dataset is organized into three main subsets:
1. **Training set**: Contains 8,219 images and annotations used for training object detection models.
2. **Validation set**: Consists of 588 images and annotations used for model validation during training.
3. **Test set**: Includes 2,936 images designed for the final evaluation of trained object detection models.
Refer to the [Dataset Structure](#dataset-structure) section for more details.
### How do I configure the SKU-110k dataset for training?
The SKU-110k dataset configuration is defined in a YAML file, which includes details about the dataset's paths, classes, and other relevant information. The `SKU-110K.yaml` file is maintained at [SKU-110K.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/SKU-110K.yaml). For example, you can train a model using this configuration as shown in our [Usage](#usage) section.
### What are the key features of the SKU-110k dataset in the context of [deep learning](https://www.ultralytics.com/glossary/deep-learning-dl)?
The SKU-110k dataset features images of store shelves from around the world, showcasing densely packed objects that pose significant challenges for object detectors:
- Over 110,000 unique SKU categories
- Diverse object appearances
- Annotations include bounding boxes and SKU category labels
These features make the SKU-110k dataset particularly valuable for training and evaluating deep learning models in object detection tasks. For more details, see the [Key Features](#key-features) section.
### How do I cite the SKU-110k dataset in my research?
If you use the SKU-110k dataset in your research or development work, please cite the following paper:
!!! quote ""
=== "BibTeX"
```bibtex
@inproceedings{goldman2019dense,
author = {Eran Goldman and Roei Herzig and Aviv Eisenschtat and Jacob Goldberger and Tal Hassner},
title = {Precise Detection in Densely Packed Scenes},
booktitle = {Proc. Conf. Comput. Vision Pattern Recognition (CVPR)},
year = {2019}
}
```
More information about the dataset can be found in the [Citations and Acknowledgments](#citations-and-acknowledgments) section.

234
algorithms/dms_yolo/code/docs/en/datasets/detect/tt100k.md Normal file
View File

@@ -0,0 +1,234 @@
---
comments: true
description: Explore the Tsinghua-Tencent 100K (TT100K) traffic sign dataset with 100,000 street view images and 30,000+ annotated traffic signs for robust detection and classification.
keywords: TT100K, Tsinghua-Tencent 100K, traffic sign detection, YOLO26, dataset, object detection, street view, traffic signs, Chinese traffic signs
---
# TT100K Dataset
The [Tsinghua-Tencent 100K (TT100K)](https://cg.cs.tsinghua.edu.cn/traffic-sign/) is a large-scale traffic sign benchmark dataset created from 100,000 Tencent Street View panoramas. This dataset is specifically designed for traffic sign detection and classification in real-world conditions, providing researchers and developers with a comprehensive resource for building robust traffic sign recognition systems.
The dataset contains **100,000 images** with over **30,000 traffic sign instances** across **221 different categories**. These images capture large variations in illuminance, weather conditions, viewing angles, and distances, making it ideal for training models that need to perform reliably in diverse real-world scenarios.
This dataset is particularly valuable for:
- Autonomous driving systems
- Advanced driver assistance systems (ADAS)
- Traffic monitoring applications
- Urban planning and traffic analysis
- Computer vision research in real-world conditions
## Key Features
The TT100K dataset provides several key advantages:
- **Scale**: 100,000 high-resolution images (2048×2048 pixels)
- **Diversity**: 221 traffic sign categories covering Chinese traffic signs
- **Real-world conditions**: Large variations in weather, illumination, and viewing angles
- **Rich annotations**: Each sign includes class label, bounding box, and pixel mask
- **Comprehensive coverage**: Includes prohibitory, warning, mandatory, and informative signs
- **Train/Test split**: Pre-defined splits for consistent evaluation
## Dataset Structure
The TT100K dataset is split into three subsets:
1. **Training Set**:
The primary collection of traffic-scene images used to train models for detecting and classifying different types of traffic signs.
2. **Validation Set**:
A subset used during model development to monitor performance and tune hyperparameters.
3. **Test Set**:
A held-out collection of images used to evaluate the final model's ability to detect and classify traffic signs in real-world scenarios.
The TT100K dataset includes 221 traffic sign categories organized into several major groups:
**Speed Limit Signs (pl*, pm*)**
1. **pl\_**: Prohibitory speed limits (pl5, pl10, pl20, pl30, pl40, pl50, pl60, pl70, pl80, pl100, pl120)
2. **pm\_**: Minimum speed limits (pm5, pm10, pm20, pm30, pm40, pm50, pm55)
**Prohibitory Signs (p*, pn*, pr\_)**
1. **p1-p28**: General prohibitory signs (no entry, no parking, no stopping, etc.)
2. **pn/pne**: No entry and no parking signs
3. **pr**: Various restriction signs (pr10, pr20, pr30, pr40, pr50, etc.)
**Warning Signs (w\_)**
1. **w1-w66**: Warning signs for various road hazards, conditions, and situations
2. Includes pedestrian crossings, sharp turns, slippery roads, animals, construction, etc.
**Height/Width Limit Signs (ph*, pb*)**
1. **ph\_**: Height limit signs (ph2, ph2.5, ph3, ph3.5, ph4, ph4.5, ph5, etc.)
2. **pb\_**: Width limit signs
**Informative Signs (i*, il*, io, ip)**
1. **i1-i15**: General informative signs
2. **il\_**: Speed limit information (il60, il80, il100, il110)
3. **io**: Other informative signs
4. **ip**: Information plates
## Dataset YAML
A YAML (Yet Another Markup Language) file is used to define the dataset configuration. It contains information about the dataset's paths, classes, and other relevant information. For the TT100K dataset, the `TT100K.yaml` file includes automatic download and conversion functionality.
!!! example "ultralytics/cfg/datasets/TT100K.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/TT100K.yaml"
```
## Usage
To train a YOLO26 model on the TT100K dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, you can use the following code snippets. The dataset will be automatically downloaded and converted to YOLO format on first use.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model - dataset will auto-download on first run
results = model.train(data="TT100K.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
# Dataset will auto-download and convert on first run
yolo detect train data=TT100K.yaml model=yolo26n.pt epochs=100 imgsz=640
```
## Sample Images and Annotations
Here are typical examples from the TT100K dataset:
1. **Urban environments**: Street scenes with multiple traffic signs at various distances
2. **Highway scenes**: High-speed road signs including speed limits and direction indicators
3. **Complex intersections**: Multiple signs in close proximity with varying orientations
4. **Challenging conditions**: Signs under different lighting (day/night), weather (rain/fog), and viewing angles
The dataset includes:
1. **Close-up signs**: Large, clearly visible signs occupying significant image area
2. **Distant signs**: Small signs requiring fine-grained detection capabilities
3. **Partially occluded signs**: Signs partially blocked by vehicles, trees, or other objects
4. **Multiple signs per image**: Images containing several different sign types
## Citations and Acknowledgments
If you use the TT100K dataset in your research or development work, please cite the following paper:
!!! quote ""
=== "BibTeX"
```bibtex
@InProceedings{Zhu_2016_CVPR,
author = {Zhu, Zhe and Liang, Dun and Zhang, Songhai and Huang, Xiaolei and Li, Baoli and Hu, Shimin},
title = {Traffic-Sign Detection and Classification in the Wild},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2016}
}
```
We would like to acknowledge the Tsinghua University and Tencent collaboration for creating and maintaining this valuable resource for the computer vision and autonomous driving communities. For more information about the TT100K dataset, visit the [official dataset website](https://cg.cs.tsinghua.edu.cn/traffic-sign/).
## FAQ
### What is the TT100K dataset used for?
The Tsinghua-Tencent 100K (TT100K) dataset is specifically designed for traffic sign detection and classification in real-world conditions. It's primarily used for:
1. Training autonomous driving perception systems
2. Developing Advanced Driver Assistance Systems (ADAS)
3. Research in robust object detection under varying conditions
4. Benchmarking traffic sign recognition algorithms
5. Testing model performance on small objects in large images
With 100,000 diverse street view images and 221 traffic sign categories, it provides a comprehensive testbed for real-world traffic sign detection.
### How many traffic sign categories are in TT100K?
The TT100K dataset contains **221 different traffic sign categories**, including:
1. **Speed limits**: pl5 through pl120 (prohibitory limits) and pm5 through pm55 (minimum speeds)
2. **Prohibitory signs**: 28+ general prohibition types (p1-p28) plus restrictions (pr\*, pn, pne)
3. **Warning signs**: 60+ warning categories (w1-w66)
4. **Height/width limits**: ph* and pb* series for physical restrictions
5. **Informative signs**: i1-i15, il\*, io, ip for guidance and information
This comprehensive coverage includes most traffic signs found in Chinese road networks.
### How can I train a YOLO26n model using the TT100K dataset?
To train a YOLO26n model on the TT100K dataset for 100 epochs with an image size of 640, use the example below.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="TT100K.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=TT100K.yaml model=yolo26n.pt epochs=100 imgsz=640
```
For detailed training configurations, refer to the [Training](../../modes/train.md) documentation.
### What makes TT100K challenging compared to other datasets?
TT100K presents several unique challenges:
1. **Scale variation**: Signs range from very small (distant highway signs) to large (close-up urban signs)
2. **Real-world conditions**: Extreme variations in lighting, weather, and viewing angles
3. **High resolution**: 2048×2048 pixel images require significant processing power
4. **Class imbalance**: Some sign types are much more common than others
5. **Dense scenes**: Multiple signs may appear in a single image
6. **Partial occlusion**: Signs may be partially blocked by vehicles, vegetation, or structures
These challenges make TT100K a valuable benchmark for developing robust detection algorithms.
### How do I handle the large image sizes in TT100K?
The TT100K dataset uses 2048×2048 pixel images, which can be resource-intensive. Here are recommended strategies:
**For Training:**
```python
# Option 1: Resize to standard YOLO size
model.train(data="TT100K.yaml", imgsz=640, batch=16)
# Option 2: Use larger size for better small object detection
model.train(data="TT100K.yaml", imgsz=1280, batch=4)
# Option 3: Multi-scale training
model.train(data="TT100K.yaml", imgsz=640, scale=0.5) # trains at varying scales
```
**Recommendations:**
- Start with `imgsz=640` for initial experiments
- Use `imgsz=1280` if you have sufficient GPU memory (24GB+)
- Consider tiling strategies for very small signs
- Use gradient accumulation to simulate larger batch sizes

179
algorithms/dms_yolo/code/docs/en/datasets/detect/visdrone.md Normal file
View File

@@ -0,0 +1,179 @@
---
comments: true
description: Explore the VisDrone Dataset, a large-scale benchmark for drone-based image and video analysis with over 2.6 million annotations for objects like pedestrians and vehicles.
keywords: VisDrone, drone dataset, computer vision, object detection, object tracking, crowd counting, machine learning, deep learning
---
# VisDrone Dataset
The [VisDrone Dataset](https://github.com/VisDrone/VisDrone-Dataset) is a large-scale benchmark created by the AISKYEYE team at the Lab of [Machine Learning](https://www.ultralytics.com/glossary/machine-learning-ml) and Data Mining, Tianjin University, China. It contains carefully annotated ground truth data for various computer vision tasks related to drone-based image and video analysis.
<p align="center">
<br>
<iframe loading="lazy" width="720" height="405" src="https://www.youtube.com/embed/9ymyH4H1fG4"
title="YouTube video player" frameborder="0"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen>
</iframe>
<br>
<strong>Watch:</strong> How to Train Ultralytics YOLO26 on the VisDrone Dataset | Aerial Detection | Complete Tutorial 🚀
</p>
VisDrone is composed of 288 video clips with 261,908 frames and 10,209 static images, captured by various drone-mounted cameras. The dataset covers a wide range of aspects, including location (14 different cities across China), environment (urban and rural), objects (pedestrians, vehicles, bicycles, etc.), and density (sparse and crowded scenes). The dataset was collected using various drone platforms under different scenarios and weather and lighting conditions. These frames are manually annotated with over 2.6 million bounding boxes of targets such as pedestrians, cars, bicycles, and tricycles. Attributes like scene visibility, object class, and occlusion are also provided for better data utilization.
## Dataset Structure
The VisDrone dataset is organized into five main subsets, each focusing on a specific task:
1. **Task 1**: Object detection in images
2. **Task 2**: Object detection in videos
3. **Task 3**: Single-object tracking
4. **Task 4**: [Multi-object tracking](../index.md#multi-object-tracking)
5. **Task 5**: Crowd counting
## Applications
The VisDrone dataset is widely used for training and evaluating deep learning models in drone-based [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv) tasks such as object detection, object tracking, and crowd counting. The dataset's diverse set of sensor data, object annotations, and attributes make it a valuable resource for researchers and practitioners in the field of drone-based computer vision.
## Dataset YAML
A YAML (Yet Another Markup Language) file is used to define the dataset configuration. It contains information about the dataset's paths, classes, and other relevant information. In the case of the Visdrone dataset, the `VisDrone.yaml` file is maintained at [https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/VisDrone.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/VisDrone.yaml).
!!! example "ultralytics/cfg/datasets/VisDrone.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/VisDrone.yaml"
```
## Usage
To train a YOLO26n model on the VisDrone dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, you can use the following code snippets. For a comprehensive list of available arguments, refer to the model [Training](../../modes/train.md) page.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="VisDrone.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=VisDrone.yaml model=yolo26n.pt epochs=100 imgsz=640
```
## Sample Data and Annotations
The VisDrone dataset contains a diverse set of images and videos captured by drone-mounted cameras. Here are some examples of data from the dataset, along with their corresponding annotations:
![VisDrone dataset aerial drone imagery with object detection](https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/visdrone-object-detection-sample.avif)
- **Task 1**: [Object detection](https://www.ultralytics.com/glossary/object-detection) in images - This image demonstrates an example of object detection in images, where objects are annotated with [bounding boxes](https://www.ultralytics.com/glossary/bounding-box). The dataset provides a wide variety of images taken from different locations, environments, and densities to facilitate the development of models for this task.
The example showcases the variety and complexity of the data in the VisDrone dataset and highlights the importance of high-quality sensor data for drone-based computer vision tasks.
## Citations and Acknowledgments
If you use the VisDrone dataset in your research or development work, please cite the following paper:
!!! quote ""
=== "BibTeX"
```bibtex
@ARTICLE{9573394,
author={Zhu, Pengfei and Wen, Longyin and Du, Dawei and Bian, Xiao and Fan, Heng and Hu, Qinghua and Ling, Haibin},
journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},
title={Detection and Tracking Meet Drones Challenge},
year={2021},
volume={},
number={},
pages={1-1},
doi={10.1109/TPAMI.2021.3119563}}
```
We would like to acknowledge the AISKYEYE team at the Lab of Machine Learning and [Data Mining](https://www.ultralytics.com/glossary/data-mining), Tianjin University, China, for creating and maintaining the VisDrone dataset as a valuable resource for the drone-based computer vision research community. For more information about the VisDrone dataset and its creators, visit the [VisDrone Dataset GitHub repository](https://github.com/VisDrone/VisDrone-Dataset).
## FAQ
### What is the VisDrone Dataset and what are its key features?
The [VisDrone Dataset](https://github.com/VisDrone/VisDrone-Dataset) is a large-scale benchmark created by the AISKYEYE team at Tianjin University, China. It is designed for various computer vision tasks related to drone-based image and video analysis. Key features include:
- **Composition**: 288 video clips with 261,908 frames and 10,209 static images.
- **Annotations**: Over 2.6 million bounding boxes for objects like pedestrians, cars, bicycles, and tricycles.
- **Diversity**: Collected across 14 cities, in urban and rural settings, under different weather and lighting conditions.
- **Tasks**: Split into five main tasks—object detection in images and videos, single-object and multi-object tracking, and crowd counting.
### How can I use the VisDrone Dataset to train a YOLO26 model with Ultralytics?
To train a YOLO26 model on the VisDrone dataset for 100 epochs with an image size of 640, you can follow these steps:
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a pretrained model
model = YOLO("yolo26n.pt")
# Train the model
results = model.train(data="VisDrone.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=VisDrone.yaml model=yolo26n.pt epochs=100 imgsz=640
```
For additional configuration options, please refer to the model [Training](../../modes/train.md) page.
### What are the main subsets of the VisDrone dataset and their applications?
The VisDrone dataset is divided into five main subsets, each tailored for a specific computer vision task:
1. **Task 1**: Object detection in images.
2. **Task 2**: Object detection in videos.
3. **Task 3**: Single-object tracking.
4. **Task 4**: Multi-object tracking.
5. **Task 5**: Crowd counting.
These subsets are widely used for training and evaluating [deep learning](https://www.ultralytics.com/glossary/deep-learning-dl) models in drone-based applications such as surveillance, traffic monitoring, and public safety.
### Where can I find the configuration file for the VisDrone dataset in Ultralytics?
The configuration file for the VisDrone dataset, `VisDrone.yaml`, can be found in the Ultralytics repository at the following link:
[VisDrone.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/VisDrone.yaml).
### How can I cite the VisDrone dataset if I use it in my research?
If you use the VisDrone dataset in your research or development work, please cite the following paper:
!!! quote ""
=== "BibTeX"
```bibtex
@ARTICLE{9573394,
author={Zhu, Pengfei and Wen, Longyin and Du, Dawei and Bian, Xiao and Fan, Heng and Hu, Qinghua and Ling, Haibin},
journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},
title={Detection and Tracking Meet Drones Challenge},
year={2021},
volume={},
number={},
pages={1-1},
doi={10.1109/TPAMI.2021.3119563}
}
```

148
algorithms/dms_yolo/code/docs/en/datasets/detect/voc.md Normal file
View File

@@ -0,0 +1,148 @@
---
comments: true
description: Discover the PASCAL VOC dataset, essential for object detection, segmentation, and classification. Learn key features, applications, and usage tips.
keywords: PASCAL VOC, VOC dataset, object detection, segmentation, classification, YOLO, Faster R-CNN, Mask R-CNN, image annotations, computer vision
---
# VOC Dataset
The [PASCAL VOC](http://host.robots.ox.ac.uk/pascal/VOC/) (Visual Object Classes) dataset is a well-known object detection, segmentation, and classification dataset. It is designed to encourage research on a wide variety of object categories and is commonly used for benchmarking computer vision models. It is an essential dataset for researchers and developers working on object detection, segmentation, and classification tasks.
<p align="center">
<br>
<iframe loading="lazy" width="720" height="405" src="https://www.youtube.com/embed/yrHzL8RyY6g"
title="YouTube video player" frameborder="0"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen>
</iframe>
<br>
<strong>Watch:</strong> How to Train Ultralytics YOLO26 on the Pascal VOC Dataset | Object Detection 🚀
</p>
## Key Features
- VOC dataset includes two main challenges: VOC2007 and VOC2012.
- The dataset comprises 20 object categories, including common objects like cars, bicycles, and animals, as well as more specific categories such as boats, sofas, and dining tables.
- Annotations include object bounding boxes and class labels for object detection and classification tasks, and segmentation masks for the segmentation tasks.
- VOC provides standardized evaluation metrics like [mean Average Precision](https://www.ultralytics.com/glossary/mean-average-precision-map) (mAP) for object detection and classification, making it suitable for comparing model performance.
## Dataset Structure
The VOC dataset is split into three subsets:
1. **Train**: This subset contains images for training object detection, segmentation, and classification models.
2. **Validation**: This subset has images used for validation purposes during model training.
3. **Test**: This subset consists of images used for testing and benchmarking the trained models. Ground truth annotations for this subset are not publicly available, and the results were historically submitted to the PASCAL VOC evaluation server for performance evaluation.
## Applications
The VOC dataset is widely used for training and evaluating [deep learning](https://www.ultralytics.com/glossary/deep-learning-dl) models in object detection (such as [Ultralytics YOLO](https://docs.ultralytics.com/models/yolo26/), [Faster R-CNN](https://arxiv.org/abs/1506.01497), and [SSD](https://arxiv.org/abs/1512.02325)), [instance segmentation](https://www.ultralytics.com/glossary/instance-segmentation) (such as [Mask R-CNN](https://arxiv.org/abs/1703.06870)), and [image classification](https://www.ultralytics.com/glossary/image-classification). The dataset's diverse set of object categories, large number of annotated images, and standardized evaluation metrics make it an essential resource for [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv) researchers and practitioners.
## Dataset YAML
A YAML (Yet Another Markup Language) file is used to define the dataset configuration. It contains information about the dataset's paths, classes, and other relevant information. In the case of the VOC dataset, the `VOC.yaml` file is maintained at [https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/VOC.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/VOC.yaml).
!!! example "ultralytics/cfg/datasets/VOC.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/VOC.yaml"
```
## Usage
To train a YOLO26n model on the VOC dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, you can use the following code snippets. For a comprehensive list of available arguments, refer to the model [Training](../../modes/train.md) page.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="VOC.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=VOC.yaml model=yolo26n.pt epochs=100 imgsz=640
```
## Sample Images and Annotations
The VOC dataset contains a diverse set of images with various object categories and complex scenes. Here are some examples of images from the dataset, along with their corresponding annotations:
![Pascal VOC dataset mosaic training batch](https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/mosaiced-voc-dataset-sample.avif)
- **Mosaiced Image**: This image demonstrates a training batch composed of mosaiced dataset images. Mosaicing is a technique used during training that combines multiple images into a single image to increase the variety of objects and scenes within each training batch. This helps improve the model's ability to generalize to different object sizes, aspect ratios, and contexts.
The example showcases the variety and complexity of the images in the VOC dataset and the benefits of using mosaicing during the training process.
## Citations and Acknowledgments
If you use the VOC dataset in your research or development work, please cite the following paper:
!!! quote ""
=== "BibTeX"
```bibtex
@misc{everingham2010pascal,
title={The PASCAL Visual Object Classes (VOC) Challenge},
author={Mark Everingham and Luc Van Gool and Christopher K. I. Williams and John Winn and Andrew Zisserman},
year={2010},
eprint={0909.5206},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
```
We would like to acknowledge the PASCAL VOC Consortium for creating and maintaining this valuable resource for the [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv) community. For more information about the VOC dataset and its creators, visit the [PASCAL VOC dataset website](http://host.robots.ox.ac.uk/pascal/VOC/).
## FAQ
### What is the PASCAL VOC dataset and why is it important for computer vision tasks?
The [PASCAL VOC](http://host.robots.ox.ac.uk/pascal/VOC/) (Visual Object Classes) dataset is a renowned benchmark for [object detection](https://www.ultralytics.com/glossary/object-detection), segmentation, and classification in computer vision. It includes comprehensive annotations like bounding boxes, class labels, and segmentation masks across 20 different object categories. Researchers use it widely to evaluate the performance of models like Faster R-CNN, YOLO, and Mask R-CNN due to its standardized evaluation metrics such as mean Average Precision (mAP).
### How do I train a YOLO26 model using the VOC dataset?
To train a YOLO26 model with the VOC dataset, you need the dataset configuration in a YAML file. Here's an example to start training a YOLO26n model for 100 epochs with an image size of 640:
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="VOC.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=VOC.yaml model=yolo26n.pt epochs=100 imgsz=640
```
### What are the primary challenges included in the VOC dataset?
The VOC dataset includes two main challenges: VOC2007 and VOC2012. These challenges test object detection, segmentation, and classification across 20 diverse object categories. Each image is meticulously annotated with bounding boxes, class labels, and segmentation masks. The challenges provide standardized metrics like mAP, facilitating the comparison and benchmarking of different computer vision models.
### How does the PASCAL VOC dataset enhance model benchmarking and evaluation?
The PASCAL VOC dataset enhances model benchmarking and evaluation through its detailed annotations and standardized metrics like mean Average [Precision](https://www.ultralytics.com/glossary/precision) (mAP). These metrics are crucial for assessing the performance of object detection and classification models. The dataset's diverse and complex images ensure comprehensive model evaluation across various real-world scenarios.
### How do I use the VOC dataset for [semantic segmentation](https://www.ultralytics.com/glossary/semantic-segmentation) in YOLO models?
To use the VOC dataset for semantic segmentation tasks with YOLO models, you need to configure the dataset properly in a YAML file. The YAML file defines paths and classes needed for training segmentation models. Check the VOC dataset YAML configuration file at [VOC.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/VOC.yaml) for detailed setups. For segmentation tasks, you would use a segmentation-specific model like `yolo26n-seg.pt` instead of the detection model.

188
algorithms/dms_yolo/code/docs/en/datasets/detect/xview.md Normal file
View File

@@ -0,0 +1,188 @@
---
comments: true
description: Explore the xView dataset, a rich resource of 1M+ object instances in high-resolution satellite imagery. Enhance detection, learning efficiency, and more.
keywords: xView dataset, overhead imagery, satellite images, object detection, high resolution, bounding boxes, computer vision, TensorFlow, PyTorch, dataset structure
---
# xView Dataset
The [xView](http://xviewdataset.org/) dataset is one of the largest publicly available datasets of overhead imagery, containing images from complex scenes around the world annotated using bounding boxes. The goal of the xView dataset is to accelerate progress in four [computer vision](https://www.ultralytics.com/glossary/computer-vision-cv) frontiers:
1. Reduce minimum resolution for detection.
2. Improve learning efficiency.
3. Enable discovery of more object classes.
4. Improve detection of fine-grained classes.
xView builds on the success of challenges like [Common Objects in Context (COCO)](../detect/coco.md) and aims to leverage computer vision to analyze the growing amount of available imagery from space in order to understand the visual world in new ways and address a range of important applications.
!!! warning "Manual Download Required"
The xView dataset is **not** automatically downloaded by Ultralytics scripts. You **must** manually download the dataset first from the official source:
- **Source:** DIUx xView 2018 Challenge by U.S. National Geospatial-Intelligence Agency (NGA)
- **URL:** [https://challenge.xviewdataset.org](https://challenge.xviewdataset.org)
**Important:** After downloading the necessary files (e.g., `train_images.tif`, `val_images.tif`, `xView_train.geojson`), you need to extract them and place them into the correct directory structure, typically expected under a `datasets/xView/` folder, **before** running the training commands provided below. Ensure the dataset is properly set up as per the challenge instructions.
## Key Features
- xView contains over 1 million object instances across 60 classes.
- The dataset has a resolution of 0.3 meters, providing higher resolution imagery than most public satellite imagery datasets.
- xView features a diverse collection of small, rare, fine-grained, and multi-type objects with [bounding box](https://www.ultralytics.com/glossary/bounding-box) annotation.
- Comes with a pretrained baseline model using the [TensorFlow](https://www.ultralytics.com/glossary/tensorflow) object detection API and an example for [PyTorch](https://www.ultralytics.com/glossary/pytorch).
## Dataset Structure
The xView dataset is composed of satellite images collected from WorldView-3 satellites at a 0.3m ground sample distance. It contains over 1 million objects across 60 classes in over 1,400 km² of imagery. The dataset is particularly valuable for [remote sensing](https://www.ultralytics.com/blog/using-computer-vision-to-analyze-satellite-imagery) applications and environmental monitoring.
## Applications
The xView dataset is widely used for training and evaluating [deep learning](https://www.ultralytics.com/glossary/deep-learning-dl) models for object detection in overhead imagery. The dataset's diverse set of object classes and high-resolution imagery make it a valuable resource for researchers and practitioners in the field of computer vision, especially for satellite imagery analysis. Applications include:
- Military and defense reconnaissance
- Urban planning and development
- Environmental monitoring
- Disaster response and assessment
- Infrastructure mapping and management
## Dataset YAML
A YAML (Yet Another Markup Language) file is used to define the dataset configuration. It contains information about the dataset's paths, classes, and other relevant information. In the case of the xView dataset, the `xView.yaml` file is maintained at [https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/xView.yaml](https://github.com/ultralytics/ultralytics/blob/main/ultralytics/cfg/datasets/xView.yaml).
!!! example "ultralytics/cfg/datasets/xView.yaml"
```yaml
--8<-- "ultralytics/cfg/datasets/xView.yaml"
```
## Usage
To train a model on the xView dataset for 100 [epochs](https://www.ultralytics.com/glossary/epoch) with an image size of 640, you can use the following code snippets. For a comprehensive list of available arguments, refer to the model [Training](../../modes/train.md) page.
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="xView.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=xView.yaml model=yolo26n.pt epochs=100 imgsz=640
```
## Sample Data and Annotations
The xView dataset contains high-resolution satellite images with a diverse set of objects annotated using bounding boxes. Here are some examples of data from the dataset, along with their corresponding annotations:
![xView dataset overhead satellite imagery with object detection](https://cdn.jsdelivr.net/gh/ultralytics/assets@main/docs/overhead-imagery-object-detection.avif)
- **Overhead Imagery**: This image demonstrates an example of [object detection](https://www.ultralytics.com/glossary/object-detection) in overhead imagery, where objects are annotated with bounding boxes. The dataset provides high-resolution satellite images to facilitate the development of models for this task.
The example showcases the variety and complexity of the data in the xView dataset and highlights the importance of high-quality satellite imagery for object detection tasks.
## Related Datasets
If you're working with satellite imagery, you might also be interested in exploring these related datasets:
- [DOTA-v2](../obb/dota-v2.md): A dataset for oriented object detection in aerial images
- [VisDrone](../detect/visdrone.md): A dataset for object detection and tracking in drone-captured imagery
- [Argoverse](../detect/argoverse.md): A dataset for autonomous driving with 3D tracking annotations
## Citations and Acknowledgments
If you use the xView dataset in your research or development work, please cite the following paper:
!!! quote ""
=== "BibTeX"
```bibtex
@misc{lam2018xview,
title={xView: Objects in Context in Overhead Imagery},
author={Darius Lam and Richard Kuzma and Kevin McGee and Samuel Dooley and Michael Laielli and Matthew Klaric and Yaroslav Bulatov and Brendan McCord},
year={2018},
eprint={1802.07856},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
```
We would like to acknowledge the [Defense Innovation Unit](https://www.diu.mil/) (DIU) and the creators of the xView dataset for their valuable contribution to the computer vision research community. For more information about the xView dataset and its creators, visit the [xView dataset website](http://xviewdataset.org/).
## FAQ
### What is the xView dataset and how does it benefit computer vision research?
The [xView](http://xviewdataset.org/) dataset is one of the largest publicly available collections of high-resolution overhead imagery, containing over 1 million object instances across 60 classes. It is designed to enhance various facets of computer vision research such as reducing the minimum resolution for detection, improving learning efficiency, discovering more object classes, and advancing fine-grained object detection.
### How can I use Ultralytics YOLO to train a model on the xView dataset?
To train a model on the xView dataset using [Ultralytics YOLO](https://docs.ultralytics.com/models/yolo26/), follow these steps:
!!! example "Train Example"
=== "Python"
```python
from ultralytics import YOLO
# Load a model
model = YOLO("yolo26n.pt") # load a pretrained model (recommended for training)
# Train the model
results = model.train(data="xView.yaml", epochs=100, imgsz=640)
```
=== "CLI"
```bash
# Start training from a pretrained *.pt model
yolo detect train data=xView.yaml model=yolo26n.pt epochs=100 imgsz=640
```
For detailed arguments and settings, refer to the model [Training](../../modes/train.md) page.
### What are the key features of the xView dataset?
The xView dataset stands out due to its comprehensive set of features:
- Over 1 million object instances across 60 distinct classes.
- High-resolution imagery at 0.3 meters.
- Diverse object types including small, rare, and fine-grained objects, all annotated with bounding boxes.
- Availability of a pretrained baseline model and examples in [TensorFlow](https://www.ultralytics.com/glossary/tensorflow) and PyTorch.
### What is the dataset structure of xView, and how is it annotated?
The xView dataset contains high-resolution satellite imagery captured by WorldView-3 satellites at a 0.3m ground sample distance, covering over 1 million objects across 60 distinct classes within approximately 1,400 km² of annotated imagery. Each object is labeled with bounding boxes, making the dataset highly suitable for training and evaluating [deep learning](https://www.ultralytics.com/glossary/deep-learning-dl) models for object detection in overhead views. For a detailed breakdown, refer to the [Dataset Structure section](#dataset-structure).
### How do I cite the xView dataset in my research?
If you utilize the xView dataset in your research, please cite the following paper:
!!! quote ""
=== "BibTeX"
```bibtex
@misc{lam2018xview,
title={xView: Objects in Context in Overhead Imagery},
author={Darius Lam and Richard Kuzma and Kevin McGee and Samuel Dooley and Michael Laielli and Matthew Klaric and Yaroslav Bulatov and Brendan McCord},
year={2018},
eprint={1802.07856},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
```
For more information about the xView dataset, visit the official [xView dataset website](http://xviewdataset.org/).