KPConv-PyTorch icon indicating copy to clipboard operation
KPConv-PyTorch copied to clipboard

test_model

Open SC-shendazt opened this issue 2 years ago • 4 comments

微信图片_20220619223653 Hi Dr.HuguesTHOMAS, There are 9 categories in DALES dataset, but the 0th category needs to be ignored, so there are only 8 categories in total. When I tested DALES dataset after training, I found such an error message, as shown in the picture, but I could still run test-Model, but when I ran miou test, All eight categories are close to zero

SC-shendazt avatar Jun 19 '22 14:06 SC-shendazt

The error is probably here: https://github.com/HuguesTHOMAS/KPConv-PyTorch/blob/3a774ff8d54a4d080fe65093b2299ede35d9735d/utils/tester.py#L433-L434

maybe change with:

 prob_names = ['_'.join(test_loader.dataset.label_to_names[label].split()) 
               for label in test_loader.dataset.label_values if label not in test_loader.dataset.ignored_labels] 

HuguesTHOMAS avatar Jun 20 '22 14:06 HuguesTHOMAS

QQ图片20220620230900 Thank you very much for your quick reply. I find that this problem will not occur if I add this code. I wonder if I am correct in modifying it this way

SC-shendazt avatar Jun 20 '22 15:06 SC-shendazt

Both strategies would work indeed

HuguesTHOMAS avatar Jun 20 '22 15:06 HuguesTHOMAS

Both strategies would work indeed

Thank you very much

SC-shendazt avatar Jun 20 '22 15:06 SC-shendazt