vividle / extractvideofeature Goto Github PK

Thank you for providing such a good repo. It helps me a lot.
I have successfully got the rgb features, but when I change the mode to flow, I met problems.
when I runned python demo_extract_i3d_feature.py -mode flow -weight_file ./models/flow_imagenet.pt
I got the error: RuntimeError: Given groups=1, weight of size [64, 2, 7, 7, 7], expected input[8, 3, 21, 229, 229] to have 2 channels, but got 3 channels instead
Can you help me out here?
Thank you again.

Hello, it's a great honor to see your feature extraction video project, this will play a big role in most of my research. But in the process of making, because you don't know the origin of the file './vids_cal.pkl' in your project, can you please provide me with the code to generate this pkl file? My mailbox is [email protected], I look forward to your answers, thank you!

Thanks for your work!
When I try to extract the flow feature, I found there is no code in the extract_features_from_video.py to use the method load_flow_batch.
By the way , Could you please explain the parameters flow x_ files_ and flow_y_files ? I don't understand their meanings. Are they folders that store X and Y optical flows respectively?