Research map / Label-efficient and robust learning / Semi-supervised learning
Semi-Supervised Domain Adaptation Based on Dual-Level Domain Mixing for Semantic Segmentation
Check 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 dual-level domain mixing and multi-teacher distillation | SSDA semantic segmentation · GTA5→Cityscapes, 100 labeled target images | 19-class mIoU (%): 54.2 -> 61.2 | Helped |
| Use dual-level domain mixing and multi-teacher distillation | SSDA semantic segmentation · SYNTHIA→Cityscapes, 100 labeled target images | 13-class mIoU (%): 62.1 -> 68.4 | Helped |
| Combine region-level and sample-level teachers for student distillation (ablation) | GTA5→Cityscapes student, 100 labeled target images · GTA5→Cityscapes, round 1 | mIoU (%): 55.9 -> 57.1 | Helped |
| Combine region-level and sample-level teachers for student distillation (ablation) | GTA5→Cityscapes student, 200 labeled target images · GTA5→Cityscapes, round 1 | mIoU (%): 56.2 -> 58.3 | Helped |
| Replace progressive updates with vanilla self-training (ablation) | GTA5→Cityscapes student, 100 labeled target images · GTA5→Cityscapes, round 4 | mIoU (%): 61.2 -> 53.3 | Hurt |
| Add image translation/style transfer (ablation) | GTA5→Cityscapes student, 100 labeled target images · GTA5→Cityscapes, round 1 | mIoU (%): 55.8 -> 57.1 | Helped |
| Add image translation/style transfer (ablation) | GTA5→Cityscapes region-level teacher, 500 labeled target images · GTA5→Cityscapes, round 1 | mIoU (%): 62.1 -> 61.7 | Hurt |
| Add image translation/style transfer (ablation) | GTA5→Cityscapes sample-level teacher · GTA5→Cityscapes, first round across labeled-data sizes | mIoU (%): 51.5 -> 53.9; 53.5 -> 54.4; 56.4 -> 58.4; 59.8 -> 61.7; 64.0 -> 65.8 | Helped |
| Add iterative progressive teacher-student training (ablation) | GTA5→Cityscapes student, 100 labeled target images · GTA5→Cityscapes, round 1 to round 4 | mIoU (%): 57.1 -> 61.2 | Helped |
| Use multi-teacher rather than a single region-level teacher (ablation) | GTA5→Cityscapes student, 500 labeled target images · GTA5→Cityscapes, round 1 | mIoU (%): 61.9 -> 62.6 | Mixed |
Closest papers by meaning
- SemiDAViL: Semi-supervised Domain Adaptation with Vision-Language Guidance for Semantic Segmentation (CVPR 2025 · similarity 0.93)
- Bidirectional Domain Mixup for Domain Adaptive Semantic Segmentation (AAAI 2023 · similarity 0.92)
- Universal Domain Adaptation for Semantic Segmentation (CVPR 2025 · similarity 0.91)
- Balanced Learning for Domain Adaptive Semantic Segmentation (ICML 2025 · similarity 0.91)
- 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)
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.