Effects of Motion-Relevant Knowledge From Unlabeled Video to Human-Object Interaction Detection.
Xue LinQi ZouXixia XuYaping HuangDing DingPublished in: IEEE Trans. Neural Networks Learn. Syst. (2023)
Keyphrases
- relevant knowledge
- space time
- spatial and temporal
- key frames
- motion features
- human actions
- video sequences
- optical flow
- video data
- video frames
- event detection
- detection algorithm
- dynamic scenes
- motion model
- knowledge base
- human motion
- motion analysis
- motion estimation
- video content
- video streams
- moving objects
- multimedia
- motion segmentation
- detection rate
- video retrieval
- motion patterns
- bounding box
- video processing
- visual features
- temporal structure