Abstract: The paucity of videos in current action classification datasets (UCF-101 and HMDB-51) has made it difficult to identify good video architectures, as most methods obtain similar performance ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback