search

ekosman / AnomalyDetectionCVPR2018-Pytorch

Pytorch version of - https://github.com/WaqasSultani/AnomalyDetectionCVPR2018
176 stars 55 forks source link

about data read #12

Closed libb999 closed 4 years ago

libb999 commented 4 years ago

thank you for your project. when i run feature_extractor.py,i get a error: TypeError: 'set' object does not support indexing

i already downloads UCF video data in '--dataset_path', c3d.pickle in '--pretrained_3d', '--annotation_path' and '--annotation_path_test' default in your project named Train_Annotation.txt and Test_Annotation.txt

PHDPeter commented 4 years ago

see a solution here (https://github.com/ekosman/AnomalyDetectionCVPR2018-Pytorch/issues/13), you need to just remove the set function from the return of vid_list

libb999 commented 4 years ago

see a solution here (#13), you need to just remove the set function from the return of vid_list

thank you very much!,but when i remove set(),i get a new error: 123

PHDPeter commented 4 years ago

I this looks like a new error unrelated to the first, maybe start a new issue and I have no good ideas for this problem and you can layout all the input/system set up you have with this problem.

PHDPeter commented 4 years ago

O wait I now have an idea what this could be :-) Thous names above are the name of weights for the C3D network. Have you downloaded the C3D network from https://github.com/DavideA/c3d-pytorch ? If so make sure you are setting the pretrained_3d to the right path, as in the c3d.pickle in the git I linked above. I hope that helps.

PHDPeter commented 4 years ago

they seem to be having the same problem here (https://github.com/ekosman/AnomalyDetectionCVPR2018-Pytorch/issues/1) so that might help

ekosman commented 4 years ago

Duplicate of #13

linchunmian commented 4 years ago

thank you for your project. when i run feature_extractor.py,i get a error: TypeError: 'set' object does not support indexing

i already downloads UCF video data in '--dataset_path', c3d.pickle in '--pretrained_3d', '--annotation_path' and '--annotation_path_test' default in your project named Train_Annotation.txt and Test_Annotation.txt

Hi, could you please provide me the UCF dataset? I cannot obtain the normal data due to the unknown corruption.

PHDPeter commented 4 years ago

The feature extracted data is in the README file if you need it but you can fix the set error you have thereby just removing the "set" part from it, detailed in this issue I think. Hope that helps.