Human pose based video compression via forward-referencing using deep learning.
S. M. A. K. RajinM. Manzur MurshedManoranjan PaulShyh Wei TengJiangang MaPublished in: VCIP (2022)
Keyphrases
- video compression
- deep learning
- human pose
- pose estimation
- human body
- action recognition
- motion compensation
- motion estimation
- body parts
- video coding
- video data
- human detection
- unsupervised learning
- machine learning
- human motion
- compression ratio
- latent variables
- motion vectors
- human actions
- computer vision
- weakly supervised
- three dimensional
- graph structure
- pairwise
- color images
- probabilistic model
- natural images
- super resolution
- text mining