Research map / Label-efficient and robust learning / Domain adaptation and generalization
Bidirectional Domain Mixup for Domain Adaptive 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 |
|---|---|---|---|
| Add BDM to ProDA | ProDA · GTA5 → Cityscapes | mIoU: 57.5 -> 63.9 (+6.4) | Helped |
| Add BDM to ProDA | ProDA · SYNTHIA → Cityscapes | mIoU: 62.0 -> 66.8 (+4.8) | Helped |
| Add bidirectional mixing rather than unidirectional mixing (ablation) | BDM · GTA5 → Cityscapes | mIoU: Ours-T2S 56.0; Ours-S2T 54.8; BDM 57.4 | Helped |
| Make DACS mixing bidirectional (ablation) | DACS · GTA5 → Cityscapes | mIoU: 50.9 -> 53.6 (+2.7) | Helped |
| Replace random cut and random paste with the proposed cut and paste (ablation) | Bidirectional DACS-style mixing · GTA5 → Cityscapes | mIoU: 53.6 -> 57.4 (+3.8) | Helped |
| Use the proposed cut with random paste (ablation) | BDM · GTA5 → Cityscapes | mIoU: 53.6 -> 53.8 | Helped |
| Use the proposed paste with random cut (ablation) | BDM · GTA5 → Cityscapes | mIoU: 53.6 -> 56.8 | Helped |
| Add class-balanced patch sampling (ablation) | BDM paste · GTA5 → Cityscapes | mIoU | Helped |
| Add spatial-continuity-based patch sampling (ablation) | BDM paste · GTA5 → Cityscapes | mIoU: +0.3 mIoU | Helped |
| Add confidence-based patch sampling (ablation) | BDM paste · GTA5 → Cityscapes | mIoU: +0.9 mIoU | Helped |
| Cut the lowest-confidence regions rather than keeping all source labels (ablation) | Source-only model · GTA5 → Cityscapes | mIoU: 36.6 -> 41.3 (+4.7) | Helped |
| Cut the highest-confidence regions (ablation) | Source-only model · GTA5 → Cityscapes | mIoU: 36.6 -> 27.8 (-8.8) | Hurt |
Closest papers by meaning
- Deliberated Domain Bridging for Domain Adaptive Semantic Segmentation (NeurIPS 2022 · similarity 0.92)
- Adapter Naturally Serves as Decoupler for Cross-Domain Few-Shot Semantic Segmentation (ICML 2025 · similarity 0.89)
- Balanced Learning for Domain Adaptive Semantic Segmentation (ICML 2025 · similarity 0.89)
- Towards Unsupervised Domain Bridging via Image Degradation in Semantic Segmentation (NeurIPS 2025 · similarity 0.89)
- Open-Set Domain Adaptation for Semantic Segmentation (CVPR 2024 · similarity 0.89)
- Universal Domain Adaptation for Semantic Segmentation (CVPR 2025 · similarity 0.88)
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.