You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Cause groundtruth of raw video is the category of action and the interval of action, should I get the frames'labels from each interval (seconds) firstly, then get the union of 6 frames' labels as a chunk's label and combine all chunk-level labels for each video?
The text was updated successfully, but these errors were encountered:
For THUMOS'14 dataset, when I get chunk-level feature vectors(e.g. Lx2048) for each video, how could I get chunk-level labels for each video?
Groundtruth of BasketballDunk_test for THUMOS'14 dataset:
Cause groundtruth of raw video is the category of action and the interval of action, should I get the frames'labels from each interval (seconds) firstly, then get the union of 6 frames' labels as a chunk's label and combine all chunk-level labels for each video?
The text was updated successfully, but these errors were encountered: