This error occurs when trying got export the model.
python3 export_yoloV8.py -w yolov8s.pt --dynamic
**/home/aiadmin/.local/lib/python3.8/site-packages/torchvision/io/image.py:13: UserWarning: Failed to load image Python extension:
warn(f"Failed to load image Python extension: {e}")**
WARNING ⚠️ 'ultralytics.yolo.v8' is deprecated since '8.0.136' and will be removed in '8.1.0'. Please use 'ultralytics.models.yolo' instead.
WARNING ⚠️ 'ultralytics.yolo.utils' is deprecated since '8.0.136' and will be removed in '8.1.0'. Please use 'ultralytics.utils' instead.
Note this warning may be related to loading older models. You can update your model to current structure with:
import torch
ckpt = torch.load("model.pt") # applies to both official and custom models
torch.save(ckpt, "updated-model.pt")
Starting: yolov8s.pt
Opening YOLOv8 model
Ultralytics YOLOv8.0.172 🚀 Python-3.8.10 torch-1.13.1 CPU (ARMv8 Processor rev 1 (v8l))
YOLOv8s summary (fused): 168 layers, 11156544 parameters, 0 gradients, 28.6 GFLOPs
Creating labels.txt file
Exporting the model to ONNX
Done: yolov8s.onnx
I have tried torch 1.9.0 and torchvision 0.10.0 and then the error goes away, but then
I get error
raise ValueError("Unsupported ONNX opset version: " + str(opset_version))
I then go back to 1.13.1 and 0.14.1 on torch and torchvision and the image error is back but the conversion seems to be ok.
Any idea what will solve the problem? Some say lower the opset and run with a different version of torch and torchvision but what effect will that have on the output?
This is solved by upgrading pytorch and torchvision to the latest.
When running the training the protobuf must be the latest too, but when doing the ONNX conversion its needed to downgrade to 3.19.1.
This error occurs when trying got export the model.
I have tried torch 1.9.0 and torchvision 0.10.0 and then the error goes away, but then I get error
raise ValueError("Unsupported ONNX opset version: " + str(opset_version))
I then go back to 1.13.1 and 0.14.1 on torch and torchvision and the image error is back but the conversion seems to be ok. Any idea what will solve the problem? Some say lower the opset and run with a different version of torch and torchvision but what effect will that have on the output?