Foreground-Background Disentanglement based on Image and Feature Co-Learning for 3D-Aware Generative Models.
Sanghyuk LeeDaeha KimByung Cheol SongPublished in: VCIP (2023)
Keyphrases
- generative model
- image features
- input image
- discriminative learning
- bayesian framework
- prior knowledge
- image segmentation
- template matching
- supervised learning
- image classification
- foreground background
- image content
- reinforcement learning
- computer vision
- learning algorithm
- multiscale
- machine learning
- probabilistic model
- active learning
- image retrieval
- object recognition
- feature points
- unsupervised learning
- image representation
- spatial information
- image matching
- bayesian networks
- similarity measure