Fully Data-Driven Pseudo Label Estimation for Pointly-Supervised Panoptic Segmentation.
Jing LiJunsong FanYuran YangShuqi MeiJun XiaoZhaoxiang ZhangPublished in: AAAI (2024)
Keyphrases
- data driven
- label field
- image labeling
- segmentation algorithm
- image segmentation
- optical flow estimation
- markov random field
- maximum a posteriori estimation
- region growing
- partial volume
- supervised learning
- fully unsupervised
- level set
- segmentation method
- learning algorithm
- segmentation accuracy
- maximum a posteriori
- image parsing
- medical images
- multiscale
- image analysis
- label fusion
- robust parameter estimation
- maximum likelihood criterion
- orientation estimation
- object detection
- feature selection
- brain mri
- robust estimation
- fully automatic
- edge detection
- optical flow