Fusing shape and motion matrices for view invariant action recognition using 3D skeletons.
Mengyuan LiuQinqin HeHong LiuPublished in: ICIP (2017)
Keyphrases
- view invariant
- action recognition
- human motion
- human motion analysis
- human actions
- shape representation
- shape analysis
- medial axis
- single view
- multi view
- arbitrary viewpoint
- viewpoint
- spatio temporal
- shape model
- shape descriptors
- motion capture
- multiple viewpoints
- invariant representation
- multiple views
- space time
- motion sequences
- video sequences
- body parts
- gesture recognition
- shape matching
- shape prior
- affine transformation
- activity recognition
- human body
- image retrieval
- image sequences