Feature Extraction

[jiaxian1-intel]

Can you please explain in more details how you did feature extraction for both ShanghaiTech and UCF datasets? It is unclear to me how you did train/test split for ShanghaiTech (some file names seem not match the original file names in ShanghaiTech). Also in the original ShanghaiTech dataset, it provides videos for training but image frames for testing. So what's the frame rate you used to extract training frames so that it matches testing set frame rate. Thanks