Research map / Label-efficient and robust learning / Domain adaptation and generalization
TACS: Taxonomy Adaptive Cross-Domain Semantic Segmentation
Rui Gong, Martin Danelljan, Dengxin Dai, Danda Pani Paudel, Ajad Chhatkuli, Fisher Yu, Luc Van Gool
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 UCT (uncertainty-rectified contrastive learning) to DACS (ablation) | DACS · SYNTHIA->Cityscapes (UDA, 16 classes) | mIoU: 48.34 -> 51.45 | Improved |
| replace CT with uncertainty-rectified UCT (ablation) | DACS+CT · SYNTHIA->Cityscapes (UDA, 16 classes) | mIoU: 49.47 -> 51.45 | Improved |
| add CT to BMS in open taxonomy (ablation) | TACS (M) · SYNTHIA->Cityscapes open, all classes | mIoU: 43.44 -> 45.86 | Improved |
| add UCT to BMS in open taxonomy (ablation) | TACS (M) · SYNTHIA->Cityscapes open, all classes | mIoU: 43.44 -> 45.99 | Improved |
| add pseudo-label relabeling RL (ablation) | TACS (M+UCT) · SYNTHIA->Cityscapes open, 6 new classes | mIoU: 16.92 -> 22.85 | Improved |
| add RL to M+UCT in open taxonomy (all classes) (ablation) | TACS (M+UCT) · SYNTHIA->Cityscapes open, all classes | mIoU: 45.99 -> 49.72 | Improved |
| add stochastic label mapping SLM to BMS (ablation) | TACS (M) · GTA5->Cityscapes coarse-to-fine, 6 fine classes | mIoU: 34.76 -> 52.28 | Improved |
| add UCT to M+SLM (ablation) | TACS (M+SLM) · GTA5->Cityscapes coarse-to-fine, all classes | mIoU: 53.68 -> 55.27 | Improved |
| add RL to M+SLM+UCT (ablation) | TACS · GTA5->Cityscapes coarse-to-fine, all classes | mIoU: 55.27 -> 55.68 | Improved |
| add SLM to baseline methods (DACS) | DACS · GTA5->Cityscapes coarse-to-fine, all classes | mIoU: 35.57 -> 18.98 | Hurt |
| add SLM to baseline methods (ADVENT) | ADVENT · GTA5->Cityscapes coarse-to-fine, all classes | mIoU: 37.25 -> 36.41 | Hurt |
| add SLM to M (ablation) | TACS (M) · Synscapes->Cityscapes implicit-overlap, 3 classes | mIoU: 47.94 -> 53.18 | Improved |
Closest papers by meaning
- Cross-Domain Semantic Segmentation on Inconsistent Taxonomy using VLMs (ECCV 2024 · similarity 0.92)
- Pulling Target to Source: A New Perspective on Domain Adaptive Semantic Segmentation (IJCV 2024 · similarity 0.91)
- DynAlign: Unsupervised Dynamic Taxonomy Alignment for Cross-Domain Segmentation (ICLR 2025 · similarity 0.91)
- Domain Adaptive Semantic Segmentation Using Weak Labels (ECCV 2020 · similarity 0.91)
- Universal Domain Adaptation for Semantic Segmentation (CVPR 2025 · similarity 0.90)
- SemiDAViL: Semi-supervised Domain Adaptation with Vision-Language Guidance for Semantic Segmentation (CVPR 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.
