Research map / Label-efficient and robust learning / Domain adaptation and generalization
LabOR: Labeling Only if Required for Domain Adaptive Semantic Segmentation
Inkyu Shin, Dong-Jin Kim, Jae Won Cho, Sanghyun Woo, Kwanyong Park, In So Kweon
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 segment-based human labels on inconsistency mask (SPL) instead of unsupervised adaptation | IAST framework · GTA5 to Cityscapes val | mIoU: 51.5 -> 66.6 | Improved |
| add point-based human labels (PPL) instead of unsupervised adaptation | IAST framework · GTA5 to Cityscapes val | mIoU: 51.5 -> 63.5 | Improved |
| use SPL segment labels instead of PPL point labels | LabOR · GTA5 to Cityscapes val | mIoU: 63.5 -> 66.6 | Improved |
| use PPL (about 40 points per image) instead of WDA point labels | LabOR · GTA5 to Cityscapes val | mIoU: 56.4 -> 63.5 | Improved |
| SPL versus fully supervised target training | SPL · GTA5 to Cityscapes val | mIoU: 66.6 vs 66.7 | No clear effect |
| add entropy minimization to SPL (Stage 3) (ablation) | SPL · GTA5 to Cityscapes val | mIoU: 66.6 -> 66.4 | Hurt |
| add entropy minimization to SPL (Stage 1) (ablation) | SPL · GTA5 to Cityscapes val | mIoU: 61.1 -> 61.5 | Improved |
| add entropy minimization to PPL (Stage 3) (ablation) | PPL · GTA5 to Cityscapes val | mIoU: 63.5 -> 63.9 | Improved |
| add entropy minimization to PPL (Stage 2) (ablation) | PPL · GTA5 to Cityscapes val | mIoU: 62.6 -> 62.3 | Hurt |
| select pixels by inconsistency mask instead of confidence or entropy | SPL · GTA5 to Cityscapes val | mIoU | Improved |
Closest papers by meaning
- Domain Adaptive Semantic Segmentation Using Weak Labels (ECCV 2020 · similarity 0.92)
- Universal Domain Adaptation for Semantic Segmentation (CVPR 2025 · similarity 0.92)
- Unsupervised Model Adaptation for Continual Semantic Segmentation (AAAI 2021 · similarity 0.91)
- Hard-Aware Instance Adaptive Self-Training for Unsupervised Cross-Domain Semantic Segmentation (TPAMI 2025 · similarity 0.91)
- Rectifying Pseudo Label Learning via Uncertainty Estimation for Domain Adaptive Semantic Segmentation (IJCV 2021 · similarity 0.91)
- Balanced Learning for Domain Adaptive Semantic Segmentation (ICML 2025 · similarity 0.90)
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.
