Source-Free Domain Adaptation for RGB-D Semantic Segmentation with Vision Transformers.
Giulia RizzoliDonald ShenajPietro ZanuttighPublished in: CoRR (2023)
Keyphrases
- domain adaptation
- semantic segmentation
- multiple sources
- target domain
- superpixels
- conditional random fields
- cross domain
- object segmentation
- transfer learning
- scene classification
- labeled data
- object categories
- training data
- semi supervised
- computer vision
- object classes
- test data
- domain specific
- sentiment classification
- text categorization
- training samples
- unlabeled data
- data points
- multiscale