Train the Yolov8 Segmentation Model with simply mask pictures

[Hydrageno]

Hello Everyone !!
I have a dataset with only mask pictures. I wannted to train the Yolov8 Segmentation Model with these masks.
And There is my training .yaml:

train: /home/test/SegmentationTest/images/train
val: /home/test/SegmentationTest/images/val
test: /home/test/SegmentationTest/images/test
# Classes
nc: 9
names: ['Background', 'Car', 'People', 'Bike', 'Curve', 'Sign', 'Fencing', 'Barricade', 'GroundRaised']
colors: [[0,0,0], [64,0,128], [64,64,0], [0,128,192], [0,0,192], [128,128,0], [64,64,128], [192,128,128], [192,64,0]]

my folder structure be like:

SegmentationTest/
├── images/
│   ├── train/
│   │   ├── image1.jpg
│   │   ├── image2.jpg
│   │   └── ...
│   └── val/
│       ├── image1.jpg
│       ├── image2.jpg
│       └── ...
└── masks/
    ├── train/
    │   ├── image1.png
    │   ├── image2.png
    │   └── ...
    └── val/
        ├── image1.png
        ├── image2.png
        └── ...

However:

FileNotFoundError: train: No labels found in /home/test/SegmentationTest/labels/train.cache

during the training process,
Looking for the solutions, thank you so much!

[UltralyticsAssistant]

👋 Hello @Hydrageno, thank you for reaching out and showing interest in Ultralytics 🚀! We recommend checking out the Docs for new users where you can explore detailed Python and CLI examples to guide you through common applications.

If this is a 🐛 Bug Report, it seems like you are encountering a FileNotFoundError. To help us assist you better, please provide a minimum reproducible example. This will assist in debugging your issue more efficiently.

If this is a custom training ❓ question, please ensure to include all the relevant information, such as sample images from your dataset and detailed training logs. Moreover, it's beneficial to review our Tips for Best Training Results.

In the meantime, if you want to join discussions or need real-time help, consider joining our Discord community 🎧. Alternatively, visit our Discourse platform for comprehensive discussions, or join our Subreddit to share insights.

Upgrade

Make sure you are using the latest version of the ultralytics package along with all requirements within a Python>=3.8 environment and PyTorch>=1.8. This helps ensure that any potential bugs you are experiencing are not already fixed in the latest version:

pip install -U ultralytics

Environments

Run YOLO in verified environments that are pre-configured with all essential dependencies like CUDA, Python, and PyTorch:

• Notebooks with free GPU:
• Google Cloud Deep Learning VM. Follow the instructions in the GCP Quickstart Guide
• Amazon Deep Learning AMI. Refer to the AWS Quickstart Guide
• Docker Image. Use the Docker Quickstart Guide

Status

Rest assured, an Ultralytics engineer will also assist you soon. If your status badge is green, it confirms that all Ultralytics CI tests are passing. CI tests check the accurate operation of all YOLO Modes and Tasks across macOS, Windows, and Ubuntu every 24 hours, and on each commit.

[Hydrageno]

I have check the environment. And there is my training.py

from ultralytics import YOLO
model = YOLO('pretrained/yolov8n-seg.pt') 
if __name__ == '__main__':
    results = model.train(data='data/training.yaml', epochs=10, imgsz=512)
    metrics = model.val()

[glenn-jocher]

@Hydrageno your script looks good for training. Ensure your dataset paths and annotations are correctly set up in training.yaml. If you encounter issues, please share the specific error message for further assistance.

[Hydrageno]

Thank you, I have solved the problem by changing the mask png to yolo format.
And there is another question.
The corresponding text does not include the "Background", should I remove the class in the data.yml just like:

names: ['Car', 'People', 'Bike', 'Curve', 'Sign', 'Fencing', 'Barricade', 'Bump']

Or remain:

names: ['BackGround', 'Car', 'People', 'Bike', 'Curve', 'Sign', 'Fencing', 'Barricade', 'Bump']

[glenn-jocher]

If "Background" is not annotated, you should remove it from the names list in your data.yaml.

[Hydrageno]

Thank you so much!!!

[glenn-jocher]

@Hydrageno you're welcome! If you have any more questions, feel free to ask.