Research map / Label-efficient and robust learning / Domain adaptation and generalization
Uncertainty Reduction for Model Adaptation in 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 |
|---|---|---|---|
| Adapt the source classifier using target pseudo-labels, entropy minimization, and feature-noise uncertainty loss | DeepLabV2 ResNet-101 · GTA5 → Cityscapes | mIoU: 45.1 vs 47.1 | Mixed |
| Adapt the source classifier using target pseudo-labels, entropy minimization, and feature-noise uncertainty loss | DeepLabV2 ResNet-101 · Synthia → Cityscapes | mIoU: 39.6 vs 43.8 | Hurt |
| Adapt the source classifier using target pseudo-labels, entropy minimization, and feature-noise uncertainty loss | DeepLabV2 ResNet-101 · Cityscapes → Cross-City Rio | mIoU: 48.9 -> 53.5 | Helped |
| Adapt the source classifier using target pseudo-labels, entropy minimization, and feature-noise uncertainty loss | DeepLabV2 ResNet-101 · Cityscapes → Cross-City Tokyo | mIoU: 47.8 -> 49.8 | Helped |
| Adapt the source classifier using target pseudo-labels, entropy minimization, and feature-noise uncertainty loss | DeepLabV2 ResNet-101 · Cityscapes → Cross-City Taipei | mIoU: 46.3 -> 50.1 | Helped |
| Adapt the source classifier using target pseudo-labels, entropy minimization, and feature-noise uncertainty loss | DeepLabV2 ResNet-101 · Cityscapes → Cross-City Rome | mIoU: 51.0 -> 53.8 | Helped |
| Use only pseudo-label loss (ablation) | Model adaptation on GTA5 → Cityscapes · GTA5 → Cityscapes | mIoU: 33.8 -> 42.24 | Helped |
| Use only entropy loss instead of pseudo-label loss (ablation) | Model adaptation on GTA5 → Cityscapes · GTA5 → Cityscapes | mIoU: 42.24 -> 19.85 | Hurt |
| Add entropy loss to pseudo-label loss (ablation) | Model adaptation on GTA5 → Cityscapes · GTA5 → Cityscapes | mIoU: 42.24 -> 42.39 | Helped |
| Add feature-noise uncertainty loss to pseudo-label loss (ablation) | Model adaptation on GTA5 → Cityscapes · GTA5 → Cityscapes | mIoU: 42.24 -> 42.72 | Helped |
| Do not freeze the main decoder weights during adaptation (ablation) | Full model adaptation method · GTA5 → Cityscapes | mIoU: 44.52 -> 45.07 | Hurt |
Closest papers by meaning
- Towards Better Stability and Adaptability: Improve Online Self-Training for Model Adaptation in Semantic Segmentation (CVPR 2023 · similarity 0.91)
- Unsupervised Model Adaptation for Continual Semantic Segmentation (AAAI 2021 · similarity 0.91)
- Source-Free Domain Adaptation for Semantic Segmentation (CVPR 2021 · similarity 0.91)
- Universal Domain Adaptation for Semantic Segmentation (CVPR 2025 · similarity 0.90)
- Open-Set Domain Adaptation for Semantic Segmentation (CVPR 2024 · similarity 0.88)
- Balanced Learning for Domain Adaptive Semantic Segmentation (ICML 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.