Update to ONNX opset 17

[glenn-jocher]

Resolves #10505

Signed-off-by: Glenn Jocher glenn.jocher@ultralytics.com

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Updated default ONNX opset version to 17 for model exporting in yolov5.

📊 Key Changes

• The default ONNX opset version has been upgraded from 12 to 17 in the export.py script.

🎯 Purpose & Impact

• 📈 Purpose: To align the exported YOLOv5 models with the latest ONNX standards, likely enabling improved performance and compatibility with modern ONNX-savvy tools and frameworks.
• 🚀 Impact: Users exporting their models to ONNX format will benefit from enhancements available in the newer opset, potentially leading to more efficient inference and broader support across various ONNX-compatible platforms.

[jelena2712]

The error after this merge iwhen I try to convert pt to onnx is:
Unsupported ONNX opset version: 17

[glenn-jocher]

@jelena2712 ah, you probably need to update onnx, i.e.:

pip install -U onnx

I suppose we should update requirements.txt to the min version supporting opset 17 now.

[glenn-jocher]

https://onnxruntime.ai/docs/reference/compatibility.html

[glenn-jocher]

@jelena2712 good news 😃! Your original issue may now be fixed ✅ in PR #10526. This PR updates onnx requirements to >=1.12.0

To receive this update:

• Git – git pull from within your yolov5/ directory or git clone https://github.com/ultralytics/yolov5 again
• PyTorch Hub – Force-reload model = torch.hub.load('ultralytics/yolov5', 'yolov5s', force_reload=True)
• Notebooks – View updated notebooks
• Docker – sudo docker pull ultralytics/yolov5:latest to update your image

Thank you for spotting this issue and informing us of the problem. Please let us know if this update resolves the issue for you, and feel free to inform us of any other issues you discover or feature requests that come to mind. Happy trainings with YOLOv5 🚀!