Disentangling Content and Motion for Text-Based Neural Video Manipulation.
Levent KaracanTolga KerimogluIsmail InanTolga BirdalErkut ErdemAykut ErdemPublished in: CoRR (2022)
Keyphrases
- multimedia
- video footage
- space time
- input video
- video sequences
- multimedia data
- object motion
- online video
- temporal filtering
- spatial and temporal
- motion analysis
- video data
- image sequences
- moving camera
- motion features
- visual cues
- shot change detection
- key frames
- successive frames
- visual data
- camera motion
- human motion
- video objects
- temporal consistency
- video frames
- motion model
- dynamic scenes
- video segments
- textual descriptions
- visual features
- motion estimation
- neural network
- motion trajectories
- moving objects
- multimedia content
- video content
- optical flow
- textual features
- video clips
- semantic information
- visual motion
- video streams
- lecture videos
- human actions
- temporal continuity
- global motion
- video summarization
- motion patterns
- surveillance videos
- single frame
- video retrieval
- video surveillance
- low frame rate
- video analysis
- layered representation
- motion capture
- temporal coherence
- event detection
- motion capture data
- reference frame
- background subtraction
- user generated
- motion planning