From Coarse to Fine: Hierarchical Multi-scale Temporal Information Modeling via Sub-group Convolution for Video Action Recognition.
Fengwen ChengHuicheng ZhengZehua LiuPublished in: IJCNN (2021)
Keyphrases
- coarse to fine
- temporal information
- action recognition
- multiscale
- spatial temporal
- video database
- human actions
- video sequences
- multiresolution
- hierarchical segmentation
- action classification
- temporal reasoning
- spatial information
- object detection
- bag of words
- computer vision
- action detection
- recognizing human actions
- image processing
- activity recognition
- temporal patterns
- human activities
- space time interest points
- dynamic programming
- image registration
- video clips
- image representation
- recognition of human actions
- temporal relations
- video shots
- spatio temporal
- image segmentation
- scale space
- multi class
- video content
- low level
- object recognition
- contextual information