Trear: Transformer-based RGB-D Egocentric Action Recognition.
Xiangyu LiYonghong HouPichao WangZhimin GaoMingliang XuWanqing LiPublished in: CoRR (2021)
Keyphrases
- action recognition
- activity recognition
- depth cameras
- human actions
- human detection
- bag of words
- human activities
- human activity recognition
- computer vision
- body parts
- spatial temporal
- static images
- recognition of human actions
- action classification
- recognizing actions
- depth images
- event detection
- action detection
- recognizing human actions
- view invariant
- depth sensors
- action recognition in videos
- video dataset
- human pose
- pose estimation
- space time interest points
- action primitives
- motion history images
- three dimensional