Semi-Supervised Semantic Segmentation via Dynamic Self-Training and Class-Balanced Curriculum.
Zhengyang FengQianyu ZhouGuangliang ChengXin TanJianping ShiLizhuang MaPublished in: CoRR (2020)
Keyphrases
- semi supervised
- semantic segmentation
- semi supervised learning
- weakly supervised
- object classes
- street scenes
- co training
- supervised methods
- training set
- superpixels
- object categories
- pairwise
- labeled data
- conditional random fields
- unlabeled data
- object recognition
- generative model
- multi view
- scene classification
- supervised learning
- object detection
- d objects
- hidden markov models
- active learning
- feature vectors