Error during IMX500 conversion

[JochenSchmitz]

Hello,
when exporting a YOLOv8 model to IMX500 format, I get the following error messages:
The export still works, but I'm not sure if the quantized model is really good.
Best regards,
Jochen

Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
WARNING:Model Compression Toolkit:Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
WARNING:Model Compression Toolkit:Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
WARNING:Model Compression Toolkit:Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
WARNING:Model Compression Toolkit:Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
WARNING:Model Compression Toolkit:Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
ERROR:Model Compression Toolkit:Found duplicate qco types!
ERROR:Model Compression Toolkit:Found duplicate qco types!
ERROR:Model Compression Toolkit:Found duplicate qco types!
ERROR:Model Compression Toolkit:Found duplicate qco types!
ERROR:Model Compression Toolkit:Found duplicate qco types!
ERROR:Model Compression Toolkit:Found duplicate qco types!
ERROR:Model Compression Toolkit:Found duplicate qco types!
ERROR:Model Compression Toolkit:Found duplicate qco types!
ERROR:Model Compression Toolkit:Found duplicate qco types!
Mixed precision enabled.
Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
WARNING:Model Compression Toolkit:Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
WARNING:Model Compression Toolkit:Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
WARNING:Model Compression Toolkit:Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
WARNING:Model Compression Toolkit:Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
WARNING:Model Compression Toolkit:Pytorch model has a parameter or constant Tensor value. This can cause unexpected behaviour when converting the model.
ERROR:Model Compression Toolkit:Found duplicate qco types!
ERROR:Model Compression Toolkit:Found duplicate qco types!
ERROR:Model Compression Toolkit:Found duplicate qco types!
ERROR:Model Compression Toolkit:Found duplicate qco types!
ERROR:Model Compression Toolkit:Found duplicate qco types!
ERROR:Model Compression Toolkit:Found duplicate qco types!
ERROR:Model Compression Toolkit:Found duplicate qco types!
ERROR:Model Compression Toolkit:Found duplicate qco types!
ERROR:Model Compression Toolkit:Found duplicate qco types!

[UltralyticsAssistant]

👋 Hello @JochenSchmitz, thank you for sharing your question and for your interest in Ultralytics 🚀! We recommend taking a look at the Ultralytics Docs for detailed guidance and examples on export-related workflows, including IMX500 export processes.

To assist further, if this is a 🐛 Bug Report, please provide us with a minimum reproducible example that demonstrates the issue. This will help us effectively debug and identify the underlying cause. For example, include your environment details, the exact ultralytics version used, and the specific export command you are running.

Upgrade

Please ensure that you are using the latest version of Ultralytics and related dependencies. You can upgrade with the following command:

pip install -U ultralytics

Recommended Environments

We also suggest testing in one of the following verified environments to rule out potential issues related to dependencies:

• Notebooks with free GPUs:
  
  
  
• Other environments such as Google Cloud, AWS, or Docker .

Community and Support

For real-time discussions and additional support, consider joining our Discord 🎧. You can also engage with our community on Discourse or participate in broader discussions on our Subreddit.

CI Status

CI tests are conducted routinely to ensure the stability of all YOLO Modes and Tasks on all major operating systems.

This is an automated response to ensure you get guidance quickly. An Ultralytics engineer will review and assist you further soon. Thanks again for contributing to the Ultralytics community ✨!

[glenn-jocher]

@JochenSchmitz hi Jochen, the warnings and errors appear to be related to the Model Compression Toolkit (MCT) during the quantization process. While the export may succeed, these issues could affect the quantized model’s quality. Please ensure you're using compatible versions of PyTorch and MCT and that your model complies with IMX500 requirements. If issues persist, consider referring to the detailed MCT quantization support in the documentation here or consult the toolkit's official resources.

[JochenSchmitz]

@glenn-jocher Hi. I am sure i follow this guideline. I can not do any more than

yolo export format=imx model=best.pt data=data.yaml imgsz=640

Maybe someone found the same issue and a solution

Regards

Jochen

[glenn-jocher]

@JochenSchmitz hi Jochen, based on your command, ensure all required dependencies are correctly installed and up to date, as specified in the IMX500 export guide. If the issue persists, please verify your model meets IMX500 constraints, such as detection-specific tasks only. Let us know if additional details arise!

[LucasSchneiderRS]

Having the same problem here.
The model .pt work OK running directly in YOLO
But, when convert I have these warnings to, and the model can´t predict the custom models
I bought the AI Camera becouse i thoungh that i could get good performance in a simple way. I was wrong hehehe

[glenn-jocher]

@LucasSchneiderRS thank you for sharing your experience. The warnings during IMX500 conversion often stem from compatibility issues or quantization constraints. Ensure your custom model adheres to the IMX500 requirements, including being detection-specific and properly quantized. You can find detailed export instructions and recommendations in the IMX500 documentation. Let us know if you encounter further issues!