Exploiting spatio-temporal representation for 3D human action recognition from depth map sequences.
Xiaopeng JiQingsong ZhaoJun ChengChenfei MaPublished in: Knowl. Based Syst. (2021)
Keyphrases
- action recognition
- spatio temporal interest points
- depth map
- action recognition in videos
- spatio temporal
- view invariant
- human actions
- multi view
- motion capture data
- recognition of human actions
- motion history images
- recognizing human actions
- image sequences
- depth information
- d scene
- depth cameras
- human activities
- three dimensional
- action classification
- activity recognition
- stereo matching
- human motion
- single view
- computer vision
- high quality
- view synthesis
- multiple views
- viewpoint
- stereo pair
- space time
- human pose
- dynamic scenes
- super resolution
- input image
- high resolution
- body parts
- disparity map
- depth estimation
- input data
- low resolution
- single image
- machine learning