Weakly supervised coarse-to-fine learning for human action segmentation in HCI videos.
Longshuai ShengCe LiPublished in: Multim. Tools Appl. (2023)
Keyphrases
- weakly supervised
- coarse to fine
- multiscale
- human actions
- multiresolution
- learning algorithm
- superpixels
- human computer interaction
- segmentation algorithm
- human motion
- image registration
- action recognition
- viewpoint
- shape prior
- video sequences
- supervised learning
- image analysis
- topic models
- segmentation method
- video frames
- human activities
- pairwise
- object class
- machine learning