Leveraging Motion Priors in Videos for Improving Human Segmentation.
Yu-Ting ChenWen-Yen ChangHai-Lun LuTingfan WuMin SunPublished in: ECCV (7) (2018)
Keyphrases
- motion capture data
- static images
- motion cues
- perceptual information
- independently moving objects
- video objects
- temporal segmentation
- human motion
- image sequences
- motion based segmentation
- motion segmentation
- object segmentation
- motion capture
- video stabilization
- rigidly moving
- video sequences
- video scene
- dynamic scenes
- multibody
- periodic motion
- level set
- optical flow
- motion parameters
- motion model
- image motion
- motion features
- video frames
- motion analysis
- camera motion
- motion estimation
- recognition of human actions
- temporal coherence
- key frames
- motion history images
- input video
- human arm
- segmentation algorithm
- moving observer
- image segmentation
- shape prior
- space time
- motion detection
- human activities
- prior model
- spatial and temporal
- moving objects
- moving camera
- foreground objects
- segmentation method
- successive frames
- optical flow estimation
- background subtraction
- spatial coherence
- point trajectories
- object motion
- prior information
- video analysis
- bayesian framework
- neighboring frames
- spatio temporal
- parametric models
- video data