Login / Signup

DTR: An Information Bottleneck Based Regularization Framework for Video Action Recognition.

Jiawei FanYu ZhaoXie YuLihua MaJunqi LiuFangqiu YiBoxun Li
Published in: ACM Multimedia (2022)
Keyphrases
  • action recognition
  • human actions
  • human activities
  • computer vision
  • video sequences
  • space time
  • video data
  • semi supervised learning
  • video frames
  • human motion
  • max margin