PoseFlow: A Deep Motion Representation for Understanding Human Behaviors in Videos.
Dingwen ZhangGuangyu GuoDong HuangJunwei HanPublished in: CVPR (2018)
Keyphrases
- human behavior
- recognition of human actions
- recognizing human actions
- video representation
- temporal domain
- action recognition in videos
- space time
- action recognition
- spatial and temporal
- video sequences
- motion features
- human actions
- image sequences
- periodic motion
- dynamic scenes
- spatio temporal
- view invariant
- moving camera
- motion analysis
- camera motion
- motion estimation
- input video
- static images
- motion model
- daily life
- video content
- human motion
- motion capture data
- video analysis
- video frames
- object recognition