Using convert_coco

[Josephts1]

Hello everyone.
I have used the labelme tool to label my dataset, this tool returns the labels in .JSON format. I have to convert a whole folder with the labels to YOLO format (.txt). I investigated and I can do it with the convert_coco function but I have not been able to solve the error shown in the image (see image). I would appreciate if you could help me to correct this problem. I am using a virtual environment in visual studio code

[UltralyticsAssistant]

👋 Hello @Josephts1, thank you for sharing your query and for using Ultralytics 🚀!

We recommend you take a look at our Docs for resources on dataset preparation and related functionalities. If you're utilizing convert_coco, make sure to explore our Python API documentation for understanding how to properly leverage available functions.

Since you're encountering an error, if this is a 🐛 Bug Report, we kindly request that you provide a minimum reproducible example (MRE) to help us debug it as efficiently as possible. Are you certain that your LabelMe JSON files comply with the expected COCO format, and that the error isn’t related to input formatting? Sharing code snippets, command details, or error logs would be greatly beneficial! 😊

Here are a few additional suggestions:

1️⃣ Ensure you're running the latest version of the ultralytics package. Upgrade your package in a Python>=3.8 environment with the following command:

pip install -U ultralytics

2️⃣ Verify you have all required dependencies installed, per the requirements. Also, confirm your virtual environment setup is properly configured as you're running this in Visual Studio Code.

Environments

If you'd like a quick and verified environment for running the convert_coco function, we recommend using one of these setups with all dependencies pre-installed:

• 
• 
• 

💡 Pro tip: You might need to preprocess LabelMe annotations into standard COCO format before utilizing convert_coco. Let us know how you're currently using the function.

If you have further questions or need guidance on custom training ❓, please include additional details such as a few sample LabelMe JSON files, any partial outputs, or screenshots of error messages.

Rest assured, this is an automated response, but an Ultralytics engineer will investigate and assist you further soon. For real-time community support, don’t hesitate to join us on Discord 🎧, dive into our Discourse, or discuss with fellow users on our Subreddit.

[Josephts1]

Figure 1 shows an example of a tag for an image in .JSON format.
I also tried to use the Google Colab tool to fix the error but it didn't work, I get another error.
Please explain in detail.

[glenn-jocher]

@Josephts1 thank you for providing additional details. The convert_coco function in the Ultralytics library expects COCO-formatted JSON annotations. Since LabelMe outputs annotations in its own format, you may need to first convert the LabelMe JSON to COCO format before using convert_coco.

I recommend using tools like labelme2coco (available on GitHub) to transform your LabelMe annotations to COCO format. Once the conversion is done, ensure the resulting COCO file structure matches the expected format, and then run convert_coco as described in the documentation.

If you encounter errors during this process, please share the exact error logs so that we can assist further.