Is demo.py supposed to work with *ONNX* models ?

[vyi]

It seems that running the demo.py script with argument --weights=weights/yolop-640-640.onnx causes a pickling error.

Even though demo.py has provision for passing it a model, it could be the case that authors didn't implement the ONNX compatibility in demo.py.

Here's the error log when I run demo.py with ONNX model args:

Traceback (most recent call last):
  File "/home/bla/Documents/2024/yoloP_local/tools/demo.py", line 180, in <module>
    detect(cfg,opt)
  File "/home/bla/Documents/2024/yoloP_local/tools/demo.py", line 55, in detect
    checkpoint = torch.load(opt.weights[0], map_location= device)
  File "/home/bla/anaconda3/lib/python3.9/site-packages/torch/serialization.py", line 608, in load
    return _legacy_load(opened_file, map_location, pickle_module, **pickle_load_args)
  File "/home/bla/anaconda3/lib/python3.9/site-packages/torch/serialization.py", line 777, in _legacy_load
    magic_number = pickle_module.load(f, **pickle_load_args)
_pickle.UnpicklingError: invalid load key, '\x08'.

System info:

• Ubuntu with a cuda supported gpu
• pytorch 1.10.1

Somebody help out with this confusion. Thanks in advance!