DECNet: Dense embedding contrast for unsupervised semantic segmentation.
Xiaoqin ZhangBaiyu ChenXiaolong ZhouSixian ChanPublished in: Neural Networks (2024)
Keyphrases
- semantic segmentation
- supervised methods
- weakly supervised
- street scenes
- superpixels
- conditional random fields
- label transfer
- scene classification
- object class
- semi supervised
- object categories
- object classes
- pascal voc
- input image
- multiscale
- feature extraction
- long range
- image understanding
- supervised learning
- object detection
- small number
- image set
- object detectors
- relation extraction
- viewpoint
- object recognition
- sift flow
- bayesian networks