Exploring Cross-Image Pixel Contrast for Semantic Segmentation.
Wenguan WangTianfei ZhouFisher YuJifeng DaiEnder KonukogluLuc Van GoolPublished in: CoRR (2021)
Keyphrases
- semantic segmentation
- image pixels
- ordering constraints
- superpixels
- conditional random fields
- scene classification
- label transfer
- weakly supervised
- object categories
- object class
- pascal voc
- bag of words
- image set
- multiscale
- pixel values
- object classes
- object segmentation
- high level
- image regions
- prior knowledge
- probabilistic model
- image data
- object detection
- image classification
- feature selection
- image understanding