MAU: A Motion-Aware Unit for Video Prediction and Beyond.
Zheng ChangXinfeng ZhangShanshe WangSiwei MaYan YeXiang XinguangWen GaoPublished in: NeurIPS (2021)
Keyphrases
- space time
- video sequences
- object motion
- input video
- temporal filtering
- key frames
- spatial and temporal
- successive frames
- moving camera
- motion estimation
- low frame rate
- prediction accuracy
- video streams
- visual cues
- video data
- motion features
- visual data
- image sequences
- motion prediction
- shot change detection
- dynamic scenes
- video frames
- video content
- temporal consistency
- video representation
- motion analysis
- surveillance videos
- motion patterns
- spatio temporal
- video footage
- prediction model
- camera motion
- temporal continuity
- multimedia
- video scene
- single frame
- real time
- static images
- motion capture
- prediction error
- motion detection
- video clips
- video signals
- video retrieval
- moving objects
- periodic motion
- human actions
- video objects
- layered representation
- image frames
- video surveillance
- dynamic textures
- optical flow
- motion capture data
- human motion
- temporal domain
- temporal coherence
- reference frame
- video summarization
- image motion
- motion field
- motion trajectories
- motion model
- video analysis
- event detection
- object tracking
- scalable video coding