TA2N: Two-Stage Action Alignment Network for Few-Shot Action Recognition.
Shuyuan LiHuabin LiuRui QianYuxi LiJohn SeeMengjuan FeiXiaoyuan YuWeiyao LinPublished in: AAAI (2022)
Keyphrases
- action recognition
- human actions
- action classification
- recognizing human actions
- recognizing actions
- action primitives
- recognition of human actions
- view invariant
- bag of words
- action detection
- human detection
- activity recognition
- spatial temporal
- video sequences
- computer vision
- space time interest points
- video shots
- visual features
- body parts
- motion history images
- motion recognition
- spatio temporal
- bag of features
- static images
- human pose
- human activities
- atomic actions
- video content
- depth sensors
- video dataset
- mid level
- action models
- temporal information
- space time
- video data
- human motion