Joint Spatio-Temporal Action Localization in Untrimmed Videos with Per-Frame Segmentation.
Xuhuan DuanLe WangChangbo ZhaiNanning ZhengQilin ZhangZhenxing NiuGang HuaPublished in: ICIP (2018)
Keyphrases
- human actions
- spatio temporal
- video frames
- temporal segmentation
- action recognition
- video representation
- temporal coherence
- spatial and temporal
- view invariant
- action classification
- video objects
- spatio temporal patterns
- neighboring frames
- video sequences
- optic disc
- recognition of human actions
- space time
- level set
- image segmentation
- image sequences
- temporal domain
- single frame
- video content
- accurate localization
- segmentation algorithm
- joint action
- recognizing human actions
- multiscale
- segmentation method
- human motion
- motion segmentation
- successive frames
- segmentation accuracy
- object segmentation
- key frames
- moving objects
- medical images
- video data
- video database
- input video
- visual features
- moving object extraction
- action recognition in videos
- frame rate
- region growing
- human activities
- activity recognition
- foreground background segmentation
- motion cues
- multi view
- video signals
- background subtraction
- video analysis
- foreground objects