search

JonasSchult / Mask3D

Mask3D predicts accurate 3D semantic instances achieving state-of-the-art on ScanNet, ScanNet200, S3DIS and STPLS3D.
MIT License
525 stars 103 forks source link

TypeError: cannot pickle 'MinkowskiConvolutionFunction' object #129

Open Linsanity1 opened 1 year ago

Linsanity1 commented 1 year ago

Hi Jonas,

When I run main_instance_segmentation.py with s3dis dataset (by modifying the defaults in config_base_instance_segmentation.yaml), the code reaches the line runner.fit(model), and an error occurs:

TypeError: print_exception() got an unexpected keyword argument 'etype'
  File "/home/workspace/Mask3D/main_instance_segmentation.py", line 108, in main train(cfg)
  File "/home/workspace/Mask3D/main_instance_segmentation.py", line 84, in train runner.fit(model)
TypeError: cannot pickle 'MinkowskiConvolutionFunction' object
During handling of the above exception, another exception occurred:
  File "/home//workspace/Mask3D/main_instance_segmentation.py", line 114, in <module> main()
TypeError: print_exception() got an unexpected keyword argument 'etype'

The error seems to be related to MinkowskiEngine, but I cannot trace where the error occurs by debugging.

Could you please advise me on how to solve this problem?

Best, Luis

Huang2002200 commented 1 month ago

I met the same problem and I found that when I changed the multi-GPU training to single GPU, the problem was solved, I hope this helps!