Discovering discriminative action parts from mid-level video representations.
Michalis RaptisIasonas KokkinosStefano SoattoPublished in: CVPR (2012)
Keyphrases
- mid level
- action recognition
- human actions
- low level
- recognizing human actions
- visual cues
- high level
- event detection
- low level features
- visual concepts
- levels of abstraction
- visual information
- perceptual grouping
- video data
- superpixels
- higher level
- video content
- video analysis
- multiscale
- bag of words
- semantic concepts
- video sequences
- spatial relations
- computer vision
- mid level features
- multimedia