Research map / Label-efficient and robust learning / Domain adaptation and generalization
Category Anchor-Guided Unsupervised Domain Adaptation for Semantic Segmentation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use category-anchor-guided adaptation rather than the strongest reported baseline | CAG-UDA · GTA5→Cityscapes val | mIoU: 48.5 -> 50.2 | Helped |
| use category-anchor-guided adaptation rather than DISE | CAG-UDA · SYNTHIA→Cityscapes val (16 categories) | mIoU: 41.5 -> 44.5 | Helped |
| use category-anchor-guided adaptation rather than BLF | CAG-UDA · SYNTHIA→Cityscapes val (13 categories) | mIoU*: 51.4 -> 52.6 | Helped |
| add category-anchor-based target CE instead of probability-based target CE (ablation) | Warm-up · GTA5→Cityscapes | mIoU: 44.3 -> 46.1 | Helped |
| add category-anchor-based source and target distance losses (ablation) | Warm-up · GTA5→Cityscapes | mIoU: 41.4 -> 42.1 | Helped |
| use category-anchor-based target CE rather than category-anchor distance losses alone (ablation) | Warm-up · GTA5→Cityscapes | mIoU: 42.1 -> 46.1 | Helped |
| combine category-anchor distance losses with category-anchor target CE (ablation) | Warm-up · GTA5→Cityscapes | mIoU: 46.1 -> 46.6 | Helped |
| add the full category-anchor-guided losses and stagewise training (ablation) | CAG-UDA · GTA5→Cityscapes | mIoU: 41.4 -> 50.2 | Helped |
| train for Stage 2 rather than Stage 1 (ablation) | CAG-UDA · GTA5→Cityscapes | mIoU: 47.2 -> 49.5 | Helped |
| train for Stage 3 rather than Stage 2 (ablation) | CAG-UDA · GTA5→Cityscapes | mIoU: 49.5 -> 50.2 | Helped |
| add an extra training stage after Stage 3 (ablation) | CAG-UDA · GTA5→Cityscapes | mIoU: 50.2 -> 50.2 | No clear effect |
| remove the global feature-alignment warm-up (ablation) | CAG-UDA · GTA5→Cityscapes | mIoU: 6.3 mIoU | Hurt |
Closest papers by meaning
- Universal Domain Adaptation for Semantic Segmentation (CVPR 2025 · similarity 0.94)
- Coarse-To-Fine Domain Adaptive Semantic Segmentation With Photometric Alignment and Category-Center Regularization (CVPR 2021 · similarity 0.92)
- Balanced Learning for Domain Adaptive Semantic Segmentation (ICML 2025 · similarity 0.92)
- DynAlign: Unsupervised Dynamic Taxonomy Alignment for Cross-Domain Segmentation (ICLR 2025 · similarity 0.91)
- Towards Unsupervised Domain Bridging via Image Degradation in Semantic Segmentation (NeurIPS 2025 · similarity 0.90)
- Open-Set Domain Adaptation for Semantic Segmentation (CVPR 2024 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.