Trajectory-aware three-stream CNN for video action recognition.
Zhengkui WengYepeng GuanPublished in: J. Electronic Imaging (2019)
Keyphrases
- action recognition
- human actions
- action classification
- spatial temporal
- video dataset
- action detection
- recognition of human actions
- static images
- motion features
- recognizing human actions
- space time interest points
- computer vision
- human activities
- bag of words
- activity recognition
- motion history images
- mid level
- video data
- video sequences
- human detection
- atomic actions
- recognizing actions
- bag of features
- body parts
- video clips
- video streams
- motion capture data
- data streams
- depth sensors
- multimedia
- view invariant
- stream processing
- video content
- key frames
- motion trajectories
- human pose
- video retrieval
- spatio temporal
- view invariant action recognition