Weakly supervised cross-view action recognition via sequential motion accumulation.
Yi LiuLei QinZhongwei ChengYanhao ZhangWeigang ZhangQingming HuangPublished in: ICIP (2014)
Keyphrases
- weakly supervised
- cross view action recognition
- object class
- superpixels
- topic models
- relation extraction
- image sequences
- multi view
- semi supervised
- space time
- named entities
- object detection
- human motion
- view invariant
- moving objects
- question answering
- action recognition
- object detectors
- target object
- feature space
- multiscale