Research map / Label-efficient and robust learning / Domain adaptation and generalization
Masked Representation Modeling for Domain-Adaptive Segmentation
Wenlve Zhou, Zhiheng Zhou, Tiantao Xian, Yikui Zhai, Weibin Wu, Biyun MA
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 MRM to DACS | DACS · GTA5->Cityscapes | mIoU: 52.1 -> 55.9 | Improved |
| add MRM to MIC | MIC · GTA5->Cityscapes | mIoU: 75.9 -> 77.5 | Improved |
| add MRM to DACS | DACS · Synthia->Cityscapes (16 classes) | mIoU: 48.3 -> 55.8 | Improved |
| add MRM to MIC | MIC · Synthia->Cityscapes (16 classes) | mIoU: 67.3 -> 68.1 | Improved |
| add rebuilding to masking (ablation) | MRM · GTA5->Cityscapes (DeepLabV2-R101, DACS, ablation) | mIoU: 51.9 -> 55.9 | Improved |
| use pixel-wise classification objective instead of pixel regression (ablation) | MRM · GTA5->Cityscapes (DeepLabV2-R101, DACS, ablation) | mIoU: 51.8 -> 55.9 | Improved |
| increase masking ratio far above 40% (ablation) | MRM · GTA5->Cityscapes (DeepLabV2-R101, DACS, ablation) | mIoU | Hurt |
| use 8 Transformer blocks instead of 2 (ablation) | MRM · GTA5->Cityscapes (DeepLabV2-R101, DACS, ablation) | mIoU: 55.9 -> 52.9 | Hurt |
| raise loss weight to 10 from 1 (ablation) | MRM · GTA5->Cityscapes (DeepLabV2-R101, DACS, ablation) | mIoU: 55.9 -> 52.1 | Hurt |
| apply MRM to target domain instead of source (ablation) | DACS · GTA5->Cityscapes | mIoU: 52.9 -> 55.9 | Improved |
| also apply MRM to source images (ablation) | DACS · GTA5->Cityscapes | mIoU: 55.9 -> 55.2 | Hurt |
| train both encoder and decoder instead of encoder only (ablation) | MRM on DACS · GTA5->Cityscapes | mIoU: 54.9 -> 55.9 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- DAFormer — DAFormer: Improving Network Architectures and Training Strategies for Domain-Adaptive Semantic Segmentation (CVPR 2022) · better on 2 of 2 shared results
- HRDA — HRDA: Context-Aware High-Resolution Domain-Adaptive Semantic Segmentation (ECCV 2022) · better on 2 of 2 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
- MICDrop: Masking Image and Depth Features via Complementary Dropout for Domain-Adaptive Semantic Segmentation (ECCV 2024 · similarity 0.90)
- MaskTwins: Dual-form Complementary Masking for Domain-Adaptive Image Segmentation (ICML 2025 · similarity 0.89)
- MaskViM: Domain Generalized Semantic Segmentation with State Space Models (AAAI 2025 · similarity 0.89)
- Towards Unsupervised Domain Bridging via Image Degradation in Semantic Segmentation (NeurIPS 2025 · similarity 0.88)
- Universal Domain Adaptation for Semantic Segmentation (CVPR 2025 · similarity 0.87)
- Open-Set Domain Adaptation for Semantic Segmentation (CVPR 2024 · similarity 0.87)
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.
