Depth-based action recognition using multiscale sub-actions depth motion maps and local auto-correlation of space-time gradients.
Chengwu LiangLin QiEnqing ChenLing GuanPublished in: BTAS (2016)
Keyphrases
- space time
- human actions
- multiscale
- spatio temporal
- video sequences
- spatial and temporal
- motion patterns
- auto correlation
- dynamic scenes
- video representation
- wavelet transform
- scan statistic
- input video
- human motion
- human activities
- action recognition
- image representation
- image processing
- interesting events
- three dimensional