Feature-Independent Action Spotting without Human Localization, Segmentation, or Frame-wise Tracking.
Chuan SunMarshall F. TappenHassan ForooshPublished in: CVPR (2014)
Keyphrases
- accurate localization
- human walking
- single frame
- moving object extraction
- motion segmentation
- optic disc
- image segmentation
- automatic initialization
- segmentation method
- compressed video sequences
- segmentation algorithm
- level set
- multiscale
- visual tracking
- occlusion reasoning
- localization method
- feature vectors
- localization algorithm
- pairwise
- shape prior
- appearance model
- video frames
- object tracking
- medical images
- particle filter
- multiple objects
- image sequences
- consecutive frames
- feature correspondences
- kalman filter
- motion model
- human gait
- temporal continuity
- frame buffer
- mobile robot localization
- computer vision
- upper body
- guide wire
- optical flow
- segmented regions
- object detection
- video objects
- complex background
- bounding box
- markov random field