Exploring more concentrated and consistent activation regions for cross-domain semantic segmentation.
Muxin LiaoGuoguang HuaShishun TianYuhang ZhangWenbin ZouXia LiPublished in: Neurocomputing (2022)
Keyphrases
- cross domain
- semantic segmentation
- superpixels
- weakly supervised
- conditional random fields
- knowledge transfer
- input image
- transfer learning
- text categorization
- scene classification
- object categories
- image regions
- object class
- image features
- object recognition
- object classes
- machine learning
- image set
- image understanding
- knn