Semi-Supervised Confidence-Level-based Contrastive Discrimination for Class-Imbalanced Semantic Segmentation.
Kangcheng LiuPublished in: CoRR (2022)
Keyphrases
- confidence level
- semantic segmentation
- semi supervised
- class imbalanced
- weakly supervised
- semi supervised learning
- class imbalance
- conditional random fields
- active learning
- superpixels
- labeled data
- pairwise
- unlabeled data
- supervised learning
- feature selection
- multi view
- generative model
- unsupervised learning
- object categories
- co training
- standard deviation
- object class
- computer vision
- image annotation
- small number
- training data