MagFormer: Hybrid Video Motion Magnification Transformer from Eulerian and Lagrangian Perspectives.
Sicheng GaoYutang FengLinlin YangXuhui LiuZichen ZhuDavid S. DoermannBaochang ZhangPublished in: BMVC (2022)
Keyphrases
- space time
- key frames
- object motion
- temporal filtering
- spatial and temporal
- input video
- low frame rate
- video content
- moving camera
- video data
- image sequences
- video sequences
- motion features
- low resolution
- visual cues
- shot change detection
- video streams
- temporal consistency
- video representation
- multimedia
- fuzzy logic
- motion trajectories
- static images
- visual data
- temporal continuity
- video footage
- motion patterns
- single frame
- video frames
- camera motion
- successive frames
- dynamic scenes
- optical flow
- motion estimation
- surveillance videos
- motion analysis
- moving objects
- video objects
- motion detection
- video analysis
- temporal coherence
- video scene
- spatio temporal
- image frames
- periodic motion
- motion model
- real time
- video retrieval
- motion field
- motion capture data
- video signals
- video surveillance
- high resolution
- computer vision
- wide area motion imagery
- frame rate
- motion capture
- reference frame
- motion segmentation
- human activities
- background subtraction
- super resolution
- multi view
- optimal solution
- neural network