Deformable VisTR: Spatio temporal deformable attention for video instance segmentation.
Sudhir YarramJialian WuPan JiYi XuJunsong YuanPublished in: CoRR (2022)
Keyphrases
- spatio temporal
- temporal segmentation
- video segmentation
- spatial and temporal
- space time
- deformable shapes
- image segmentation
- video sequences
- spatial temporal
- segmentation algorithm
- deformable models
- human actions
- level set
- moving objects
- multiscale
- segmentation method
- medical images
- multimedia
- dynamic textures
- object segmentation and tracking
- shape prior
- visual attention
- image sequences
- video data
- key frames
- video database
- event detection
- background subtraction
- motion trajectories
- video scene
- temporal consistency
- video frames
- video analysis
- video representation
- deformable template
- spatio temporally
- objects in video sequences
- video content