mmpretrain: [Bug] Run demo error

Branch

main branch (mmpretrain version)

Describe the bug

I installed the mmpretrain according to the documentation, but failed in Verify the installation section.

The following is the error message (copied from terminal).

~/Documents/Fau/Projects.nosync/Python_Projects/mmpretrain (main) » python demo/image_demo.py demo/demo.JPEG resnet18_8xb32_in1k --device cpu
Loads checkpoint by http backend from path: https://download.openmmlab.com/mmclassification/v0/resnet/resnet18_8xb32_in1k_20210831-fbbb1da6.pth
Inference ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Traceback (most recent call last):
  File "/Users/Fau/Documents/Fau/Projects.nosync/Python_Projects/mmpretrain/demo/image_demo.py", line 44, in <module>
    main()
  File "/Users/Fau/Documents/Fau/Projects.nosync/Python_Projects/mmpretrain/demo/image_demo.py", line 37, in main
    result = inferencer(args.img, show=args.show, show_dir=args.show_dir)[0]
  File "/Users/Fau/Documents/Fau/Projects.nosync/Python_Projects/mmpretrain/mmpretrain/apis/image_classification.py", line 123, in __call__
    return super().__call__(inputs, return_datasamples, batch_size,
  File "/usr/local/Caskroom/miniconda/base/envs/dl/lib/python3.10/site-packages/mmengine/infer/infer.py", line 221, in __call__
    results = self.postprocess(preds, visualization, return_datasamples,
  File "/Users/Fau/Documents/Fau/Projects.nosync/Python_Projects/mmpretrain/mmpretrain/apis/image_classification.py", line 221, in postprocess
    result['pred_class'] = self.classes[pred_label]
IndexError: list index out of range

Environment

{'sys.platform': 'darwin',
 'Python': '3.10.10 (main, Mar 21 2023, 13:41:39) [Clang 14.0.6 ]',
 'CUDA available': False,
 'numpy_random_seed': 2147483648,
 'GCC': 'Homebrew clang version 16.0.2',
 'PyTorch': '2.0.0',
 'TorchVision': '0.15.1',
 'OpenCV': '4.7.0',
 'MMEngine': '0.7.3',
 'MMCV': '2.0.0',
 'MMPreTrain': '1.0.0rc7+e80418a'}

Other information

MacBook Pro 16-inch (Intel) MacOS 13.3.1a

About this issue

  • Original URL
  • State: closed
  • Created a year ago
  • Comments: 26

Most upvoted comments

There has been the same issue in PyTorch refer to https://github.com/pytorch/pytorch/issues/92311.

PyTorch would fix it in the next version.