RubiksNet: Learnable 3D-Shift for Efficient Video Action Recognition.
Linxi FanShyamal BuchGuanzhi WangRyan CaoYuke ZhuJuan Carlos NieblesLi Fei-FeiPublished in: ECCV (19) (2020)
Keyphrases
- action recognition
- human actions
- action classification
- spatial temporal
- video dataset
- recognizing human actions
- action detection
- recognition of human actions
- static images
- activity recognition
- human activities
- bag of words
- motion features
- mid level
- computer vision
- human detection
- space time interest points
- body parts
- video data
- view invariant
- bag of features
- video sequences
- recognizing actions
- motion capture data
- event recognition
- video content
- video streams
- motion history images
- action primitives
- video frames
- space time
- object detection
- spatio temporal
- view invariant action recognition