Region-Aware Metric Learning for Open World Semantic Segmentation via Meta-Channel Aggregation.
Hexin DongZifan ChenMingze YuanYutong XieJie ZhaoFei YuBin DongLi ZhangPublished in: CoRR (2022)
Keyphrases
- metric learning
- semantic segmentation
- distance metric
- superpixels
- conditional random fields
- pairwise
- semi supervised
- weakly supervised
- distance function
- object categories
- input image
- feature space
- dimensionality reduction
- object classes
- semi supervised learning
- natural images
- database
- object segmentation
- labeled data
- image set
- topic models
- unsupervised learning
- generative model
- image features
- natural language
- knowledge base
- feature selection