Render In-between: Motion Guided Video Synthesis for Action Interpolation.
Hsuan-I HoXu ChenJie SongOtmar HilligesPublished in: CoRR (2021)
Keyphrases
- recognizing human actions
- space time
- human actions
- recognition of human actions
- action recognition
- temporal domain
- human motion
- spatial and temporal
- object motion
- motion features
- input video
- key frames
- action classification
- body motions
- motion synthesis
- action detection
- successive frames
- video sequences
- temporal continuity
- moving camera
- visual cues
- video data
- temporal filtering
- static images
- low frame rate
- video footage
- image sequences
- shot change detection
- visual data
- motion patterns
- motion analysis
- motion history images
- dynamic textures
- temporal consistency
- temporal coherence
- video objects
- camera movement
- motion segmentation
- video frames
- spatio temporal
- motion capture data
- displacement estimation
- surveillance videos
- image interpolation
- motion estimation
- motion model
- multimedia
- video shots
- single frame
- video retrieval
- wide area motion imagery
- activity recognition
- video streams
- dynamic scenes
- video analysis
- motion capture
- image frames
- motion field
- video content
- motion trajectories
- camera motion
- video summarization
- layered representation
- moving objects
- global motion