G3AN: Disentangling Appearance and Motion for Video Generation.
Yaohui WangPiotr BilinskiFrançois BrémondAntitza DantchevaPublished in: CVPR (2020)
Keyphrases
- object motion
- static images
- dynamic textures
- space time
- appearance features
- layered representation
- motion cues
- video sequences
- key frames
- spatial and temporal
- image sequences
- moving camera
- temporal filtering
- video data
- input video
- successive frames
- motion model
- temporal coherence
- motion features
- dynamic scenes
- motion estimation
- low frame rate
- moving objects
- camera motion
- shot change detection
- articulated human motion
- temporal continuity
- motion trajectories
- visual cues
- visual data
- motion capture
- appearance model
- human motion
- video segmentation
- video frames
- video content
- video scene
- video summarization
- video objects
- video analysis
- optical flow
- surveillance videos
- video footage
- visual input
- motion segmentation
- single frame
- motion patterns
- motion analysis
- video signals
- motion parameters
- video surveillance
- target object
- human actions
- image frames
- camera movement
- video streams
- video clips
- action recognition