Few-Shot Semantic Segmentation based on Detail-Preserving-Aware Loss.
Chih-Chung HsuSin-Di MaPublished in: ICCE-TW (2022)
Keyphrases
- semantic segmentation
- street scenes
- superpixels
- weakly supervised
- conditional random fields
- label transfer
- scene classification
- object class
- object categories
- video sequences
- image understanding
- object classes
- multiscale
- pascal voc
- video data
- visual features
- pixel level
- input image
- probabilistic model
- image classification
- graphical models
- active learning
- object recognition