Learning Self-Similarity in Space and Time as Generalized Motion for Video Action Recognition.
Heeseung KwonManjin KimSuha KwakMinsu ChoPublished in: ICCV (2021)
Keyphrases
- action recognition
- recognition of human actions
- static images
- human actions
- space time
- recognizing human actions
- spatial temporal
- action classification
- motion features
- human movements
- spatial and temporal
- action recognition in videos
- video sequences
- view invariant
- activity recognition
- video data
- human detection
- mid level
- video dataset
- action primitives
- motion estimation
- motion patterns
- motion analysis
- key frames
- human activities
- motion history images
- bag of words