SSSC-AM: A unified framework for video co-segmentation by structured sparse subspace clustering with appearance and motion features.
Junlin YaoFrank NielsenPublished in: ICIP (2016)
Keyphrases
- motion features
- motion cues
- subspace clustering
- bounding box
- video data
- high dimensional
- motion analysis
- activity recognition
- high dimensional data
- video shots
- action recognition
- visual features
- clustering method
- motion patterns
- optical flow
- clustering algorithm
- color features
- shape features
- motion vectors
- motion model
- flow field
- video streams
- human actions
- image segmentation
- appearance model
- video content
- image sequences
- video sequences
- data sets
- depth images
- video database
- object segmentation
- video retrieval
- image classification
- dimensionality reduction
- motion segmentation
- low dimensional
- test images
- sparse representation
- segmentation method