Research map / Label-efficient and robust learning / Domain adaptation and generalization
MetaCorrection: Domain-Aware Meta Loss Correction for Unsupervised Domain 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 |
|---|---|---|---|
| Add DMLC supervision at a low-level feature layer (ablation) | Single DMLC · GTA5 → CityScapes | mIoU: 51.2 -> 52.1 | Helped |
| Add multi-level DMLC rather than using output-layer DMLC alone (ablation) | DMLC · SYNTHIA → CityScapes | mIoU: 44.6 -> 45.1 | Helped |
| Add multi-level DMLC rather than using output-layer DMLC alone (ablation) | DMLC · SYNTHIA → CityScapes | mIoU* (13 classes): 52.1 -> 52.5 | Helped |
| Add multi-level MetaCorrection to single DMLC (ablation) | Decathlon → NCI-ISBI13 · Decathlon → NCI-ISBI13 | PZ Dice: 42.03 -> 43.25 | Helped |
| Add multi-level MetaCorrection to single DMLC (ablation) | Decathlon → NCI-ISBI13 · Decathlon → NCI-ISBI13 | TZ Dice: 74.09 -> 74.31 | Helped |
| Add multi-level MetaCorrection to single DMLC (ablation) | Decathlon → NCI-ISBI13 · Decathlon → NCI-ISBI13 | WP Dice: 69.38 -> 70.87 | Helped |
| Use MetaCorrection instead of AdaptSegNet-generated pseudo labels without correction | AdaptSegNet pseudo-label self-training · GTA5 → CityScapes | mIoU: 42.4 -> 47.3 | Helped |
| Use MetaCorrection instead of LTIR-generated pseudo labels without correction | LTIR pseudo-label self-training · GTA5 → CityScapes | mIoU: 50.2 -> 52.1 | Helped |
| Use MetaCorrection instead of Source-only-generated pseudo labels without correction | Source-only pseudo-label self-training · GTA5 → CityScapes | mIoU: 36.6 -> 44.5 | Helped |
| Use MetaCorrection instead of uncertainty-based pseudo-label rectification | AdaptSegNet pseudo-label self-training · GTA5 → CityScapes | mIoU: 46.1 -> 47.3 | Helped |
Closest papers by meaning
- Continuous Pseudo-Label Rectified Domain Adaptive Semantic Segmentation With Implicit Neural Representations (CVPR 2023 · similarity 0.91)
- Universal Domain Adaptation for Semantic Segmentation (CVPR 2025 · similarity 0.91)
- Unsupervised Model Adaptation for Continual Semantic Segmentation (AAAI 2021 · similarity 0.91)
- Balanced Learning for Domain Adaptive Semantic Segmentation (ICML 2025 · similarity 0.90)
- Category Anchor-Guided Unsupervised Domain Adaptation for Semantic Segmentation (NeurIPS 2019 · similarity 0.90)
- 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.