DANNet: A One-Stage Domain Adaptation Network for Unsupervised Nighttime Semantic Segmentation.
Xinyi WuZhenyao WuHao GuoLili JuSong WangPublished in: CoRR (2021)
Keyphrases
- domain adaptation
- semantic segmentation
- weakly supervised
- semi supervised
- cross domain
- multiple sources
- labeled data
- semi supervised learning
- transfer learning
- conditional random fields
- superpixels
- sentiment classification
- supervised learning
- target domain
- unsupervised learning
- unlabeled data
- text categorization
- topic models
- wordnet
- scene classification
- computer vision