Virtual Multi-Modality Self-Supervised Foreground Matting for Human-Object Interaction.
Bo XuHan HuangCheng LuZiwen LiYandong GuoPublished in: ICCV (2021)
Keyphrases
- multi modality
- multi modal
- alpha matting
- foreground and background
- natural image matting
- medical images
- information theoretic
- image registration
- mutual information
- alpha matte
- foreground objects
- high quality
- closed form
- background subtraction
- human object interactions
- imaging modalities
- action recognition
- image sequences
- anatomical structures
- human pose
- human actions
- feature selection
- mean shift
- moving objects