Real-Time Spatio-Temporal Action Localization via Learning Motion Representation.
Yuanzhong LiuZhigang TuLiyu LinXing XieQianqing QinPublished in: ACCV Workshops (2020)
Keyphrases
- spatio temporal
- real time
- learning process
- space time
- recognition of human actions
- learning systems
- spatial and temporal
- optical flow
- active learning
- vision system
- image sequences
- relative position
- learning algorithm
- reinforcement learning
- view invariant
- video representation
- multiple representations
- position information
- surveillance videos
- motion capture
- moving target
- human actions
- motion analysis
- learning tasks
- motion model
- image representation
- visual features
- online learning
- control system
- prior knowledge