TransFGU: A Top-down Approach to Fine-Grained Unsupervised Semantic Segmentation.
Zhaoyuan YinPichao WangFan WangXianzhe XuHanling ZhangHao LiRong JinPublished in: CoRR (2021)
Keyphrases
- fine grained
- semantic segmentation
- supervised methods
- weakly supervised
- coarse grained
- conditional random fields
- superpixels
- label transfer
- scene classification
- object categories
- object class
- pascal voc
- access control
- object classes
- object detection
- topic models
- semi supervised
- image understanding
- object recognition
- image set
- multiscale
- natural scenes
- graphical models