DegardinBruno / human-self-learning-anomaly

Code for the paper "Human Activity Analysis: Iterative Weak/Self-Supervised Learning Frameworks for Detecting Abnormal Events", IJCB 2020
Other
42 stars 13 forks source link

run inference and data prepare #4

Closed carolchenyx closed 3 years ago

carolchenyx commented 3 years ago

Thanks for your excellent work. I would like to know how to run the inference.py? Where I can use the 'root_C3D_dir'? Thank you

DegardinBruno commented 3 years ago

@carolchenyx Thank you! We use C3D as our backbone network. The inference.py file will extract the C3D features directly from your video input. The "root_C3D_dir" contains the root directory for C3Dv1.0 in (https://github.com/facebookarchive/C3D). Have you already installed C3D? (Note that we only use its first version: C3Dv1.0, and not 1.1)

carolchenyx commented 3 years ago

Dear Bruno, Thanks for your quick reply. Now I know how to get the feature, but I failed to install the C3D. I would like to know do you have some methods to install it under the CUDA10.2? Thank you.

Bruno Degardin @.***> 于2021年4月12日周一 下午6:31写道:

@carolchenyx https://github.com/carolchenyx Thank you! We use C3D as our backbone network. The inference.py file will extract the C3D features directly from your video input. The "root_C3D_dir" contains the root directory for C3Dv1.0 in (https://github.com/facebookarchive/C3D). Have you already installed C3D? (Note that we only use its first version: C3Dv1.0, and not 1.1)

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/DegardinBruno/human_self_learning_anomaly/issues/4#issuecomment-817695185, or unsubscribe https://github.com/notifications/unsubscribe-auth/AL2PSUHVAY2J2VEPBHD6GOLTILDXHANCNFSM42Y5SP7A .

DegardinBruno commented 3 years ago

I understand! Its installation is really tricky... Usually, we install CUDA first, then cudnn, and finally caffe. If I remember correctly I installed caffe with (https://caffe.berkeleyvision.org/install_apt.html) CUDA and cudnn were installed through the NVIDIA tutorials. Then I installed C3D with the short tutorial here. Screenshot from 2021-04-13 12-07-45

carolchenyx commented 3 years ago

Thank you for the detailed explanation. I think I need to install this project in another computer in case of version conflicts. Super thanks again.

Bruno Degardin @.***> 于2021年4月13日周二 下午7:08写道:

I understand! Its installation is really tricky... Usually, we install CUDA first, then cudnn, and finally caffe. If I remember correctly I installed caffe with ( https://caffe.berkeleyvision.org/install_apt.html) CUDA and cudnn were installed through the NVIDIA tutorials. Then I installed C3D with the short tutorial here. [image: Screenshot from 2021-04-13 12-07-45] https://user-images.githubusercontent.com/61987997/114543268-e4826680-9c50-11eb-9d15-26fd23006ace.png

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/DegardinBruno/human_self_learning_anomaly/issues/4#issuecomment-818651858, or unsubscribe https://github.com/notifications/unsubscribe-auth/AL2PSUCUFUUN5CVUSSFYYC3TIQQ4PANCNFSM42Y5SP7A .

DegardinBruno commented 3 years ago

In case you want to try, NVIDIA graphics allow different CUDA versions installed, 10.1 (which is from this project) and 10.2, for example. Any additional information, just ask! Thank you for your interest.

carolchenyx commented 3 years ago

Thanks DegardinBruno for your useful information^_^