Describing actions in videos

Models

vid2seq