Open-World Semantic Segmentation via Contrasting and Clustering Vision-Language Embedding.
Quande LiuYoupeng WenJianhua HanChunjing XuHang XuXiaodan LiangPublished in: ECCV (20) (2022)
Keyphrases
- semantic segmentation
- closed world
- open world
- superpixels
- object categories
- data exchange
- scene classification
- incomplete information
- conditional random fields
- natural language
- object class
- weakly supervised
- data points
- image processing
- computer vision
- co occurrence
- image features
- databases
- bayesian networks
- object classes
- metadata
- pascal voc
- information retrieval
- machine learning