Virtual Multi-Modality Self-Supervised Foreground Matting for Human-Object Interaction.
Bo XuHan HuangCheng LuZiwen LiYandong GuoPublished in: CoRR (2021)
Keyphrases
- multi modality
- multi modal
- alpha matting
- foreground and background
- medical images
- natural image matting
- information theoretic
- image registration
- mutual information
- alpha matte
- imaging modalities
- high quality
- closed form
- image sequences
- foreground objects
- human pose
- background subtraction
- action recognition
- human object interactions
- low contrast
- multiresolution
- point correspondences