PointDMIG: a dynamic motion-informed graph neural network for 3D action recognition.
Yao DuZhenjie HouXing LiJiuzhen LiangKaijun YouXinwen ZhouPublished in: Multim. Syst. (2024)
Keyphrases
- action recognition
- recognizing human actions
- recognition of human actions
- static images
- human movements
- action recognition in videos
- motion features
- action primitives
- human actions
- view invariant
- mid level
- human detection
- bag of words
- activity recognition
- computer vision
- motion history images
- motion capture data
- motion analysis
- spatial temporal
- action classification
- recognizing actions
- body parts
- motion model
- space time
- action detection
- view invariant action recognition
- image sequences
- human activities
- motion estimation
- human pose
- bag of features
- graph structure
- human motion
- motion field
- optical flow
- motion capture
- motion recognition
- depth sensors
- object detection
- space time interest points
- spatio temporal