Controllable Video Generation Through Global and Local Motion Dynamics.
Aram DavtyanPaolo FavaroPublished in: ECCV (17) (2022)
Keyphrases
- space time
- articulated human motion
- spatial and temporal
- successive frames
- object motion
- video sequences
- global motion
- motion features
- temporal filtering
- input video
- motion capture
- video data
- image sequences
- human motion
- key frames
- moving camera
- visual cues
- video frames
- camera motion
- motion estimation
- video representation
- physical constraints
- motion model
- static images
- low frame rate
- spatio temporal
- multimedia
- temporal continuity
- periodic motion
- video objects
- shot change detection
- motion analysis
- dynamic scenes
- video streams
- video surveillance
- temporal consistency
- temporal coherence
- motion detection
- motion trajectories
- reference frame
- action recognition
- optical flow
- multiview video
- layered representation
- video content
- moving objects
- real time
- video footage
- wide area motion imagery
- single frame
- image motion
- video processing
- motion field
- dynamical model
- temporal domain
- dynamic model
- stationary camera
- video analysis
- image frames
- dynamic textures
- video shots
- motion capture data
- motion parameters
- humanoid robot