Online action recognition using covariance of shape and motion.
Igor KviatkovskyEhud RivlinIlan ShimshoniPublished in: Comput. Vis. Image Underst. (2014)
Keyphrases
- periodic motion
- articulated motion
- image sequences
- recognizing human actions
- motion analysis
- camera motion
- recovering shape
- camera images
- motion field
- human motion
- video objects
- optical flow
- body motions
- space time
- object shape
- motion model
- shape analysis
- shape features
- deformable surface
- motion segmentation
- shape descriptors
- motion estimation
- view invariant
- action recognition
- online learning
- geometrical constraints
- image measurements
- shape matching
- appearance features
- global motion
- motion tracking
- human actions
- nonrigid shape
- video sequences
- multi view
- graph cuts
- vector field
- covariance matrix
- shape prior
- deformable objects
- shape estimation
- shape parameters
- articulated objects
- motion detection
- image segmentation
- velocity field
- precise localization
- intensity information
- face recognition