Research map / Label-efficient and robust learning / Domain adaptation and generalization
HRDA: Context-Aware High-Resolution Domain-Adaptive Semantic Segmentation
Lukas Hoyer, Dengxin Dai, Luc Van Gool
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add HRDA to DACS with DeepLabV2 | DACS · GTA5->Cityscapes | mIoU: 53.9 -> 59.4 | Improved |
| add HRDA to DAFormer with DAFormer network | DAFormer · GTA5->Cityscapes | mIoU: 68.3 -> 73.8 | Improved |
| add HRDA to entropy minimisation | Entropy Min. DeepLabV2 · GTA5->Cityscapes | mIoU: 44.3 -> 46.7 | Improved |
| add detail HR crop to LR context crop (ablation) | HRDA · GTA5->Cityscapes | mIoU: 68.5 -> 73.8 | Improved |
| add small HR detail crop 0.25 to LR context crop (ablation) | HRDA · GTA5->Cityscapes | mIoU: 68.5 -> 70.6 | Improved |
| increase context crop size from LR 0.5 to LR 1.0 (ablation) | HRDA · GTA5->Cityscapes | mIoU: 68.5 -> 73.8 | Improved |
| use learned scale attention instead of average fusion (ablation) | HRDA · GTA5->Cityscapes | mIoU: 67.5 -> 71.5 | Improved |
| add overlapping sliding-window detail pseudo-labels (ablation) | HRDA · GTA5->Cityscapes | mIoU: 71.5 -> 72.4 | Improved |
| add detail-crop supervision loss (lambda_d=0.1) (ablation) | HRDA · GTA5->Cityscapes | mIoU: 72.4 -> 73.8 | Improved |
| use real HR detail crop instead of LR or upsampled LR crop (ablation) | HRDA · GTA5->Cityscapes | mIoU: 69.1 -> 73.8 | Improved |
| use HRDA multi-resolution crops instead of naive HR 0.75 crop | DAFormer · GTA5->Cityscapes | mIoU: 70.0 -> 73.8 | Improved |
| use context downscale 2 instead of 4 (ablation) | HRDA · GTA5->Cityscapes | mIoU: 65.9 -> 68.5 | Improved |
Later papers that compared with this method
29 later accepted papers report HRDA as a baseline in their results tables; 28 of them report a better number on most of the results they share with it.
- Masked Representation Modeling for Domain-Adaptive Segmentation (CVPR 2026 · better on 2 of 2 shared results)
- Heuristic Self-Paced Learning for Domain Adaptive Semantic Segmentation under Adverse Conditions (CVPR 2026 · better on 2 of 2 shared results)
- Rein++: Efficient Generalization and Adaptation for Semantic Segmentation With Vision Foundation Models (TPAMI 2026 · better on 1 of 1 shared result)
- Condition-Invariant Semantic Segmentation (TPAMI 2025 · better on 8 of 9 shared results)
- Balanced Learning for Domain Adaptive Semantic Segmentation (ICML 2025 · better on 3 of 3 shared results)
- Towards Unsupervised Domain Bridging via Image Degradation in Semantic Segmentation (NeurIPS 2025 · better on 3 of 3 shared results)
- Exploring Weather-aware Aggregation and Adaptation for Semantic Segmentation under Adverse Conditions (ICCV 2025 · better on 3 of 3 shared results)
- DynAlign: Unsupervised Dynamic Taxonomy Alignment for Cross-Domain Segmentation (ICLR 2025 · better on 1 of 1 shared result)
21 more not listed.
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- DAFormer — DAFormer: Improving Network Architectures and Training Strategies for Domain-Adaptive Semantic Segmentation (CVPR 2022) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Domain Adaptive and Generalizable Network Architectures and Training Strategies for Semantic Image Segmentation (TPAMI 2023 · similarity 0.93)
- Unified Domain Adaptive Semantic Segmentation (TPAMI 2025 · similarity 0.91)
- Universal Domain Adaptation for Semantic Segmentation (CVPR 2025 · similarity 0.91)
- Balanced Learning for Domain Adaptive Semantic Segmentation (ICML 2025 · similarity 0.89)
- Open-Set Domain Adaptation for Semantic Segmentation (CVPR 2024 · similarity 0.89)
- Towards Unsupervised Domain Bridging via Image Degradation in Semantic Segmentation (NeurIPS 2025 · similarity 0.89)
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.
