Research map / Label-efficient and robust learning / Domain adaptation and generalization
Divide-and-Conquer Decoupled Network for Cross-Domain Few-Shot Segmentation
Runmin Cong, Anpeng Wang, Bin Wan, Cong Zhang, Xiaofei Zhou, Wei Zhang
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 MGDF module to SSP+IFA baseline (ablation) | DCDNet baseline · FSS-1000 (mIoU) | mIoU: 80.1 -> 80.6 | Improved |
| add ACFD module before MGDF (ablation) | DCDNet (MGDF) · FSS-1000 (mIoU) | mIoU: 80.6 -> 81.3 | Improved |
| add CAM module during fine-tuning (ablation) | DCDNet (MGDF+ACFD) · FSS-1000 (mIoU) | mIoU: 81.3 -> 81.7 | Improved |
| decouple features into private and shared instead of base only (ablation) | DCDNet fusion · FSS-1000 (mIoU) | mIoU: 80.6 -> 81.4 | Improved |
| keep base feature in fusion with private and shared (ablation) | DCDNet fusion · FSS-1000 (mIoU) | mIoU: 81.4 -> 81.7 | Improved |
| add adversarial loss alone (ablation) | DCDNet losses · FSS-1000 (mIoU) | mIoU: 80.6 -> 81.0 | Improved |
| add contrastive loss alone (ablation) | DCDNet losses · FSS-1000 (mIoU) | mIoU: 80.6 -> 81.1 | Improved |
| combine adversarial and contrastive losses (ablation) | DCDNet losses · FSS-1000 (mIoU) | mIoU: 80.6 -> 81.5 | Improved |
| add orthogonality loss on top of adv+cont (ablation) | DCDNet losses · FSS-1000 (mIoU) | mIoU: 81.5 -> 81.7 | Improved |
| use DCDNet instead of IFA on ISIC 1-shot | CD-FSS ResNet-50 · ISIC 1-shot | mIoU: 66.3 -> 72.0 | Improved |
| use DCDNet instead of ABCDFSS on Chest X-Ray 5-shot | CD-FSS ResNet-50 · Chest X-Ray 5-shot | mIoU: 81.4 -> 81.1 | Hurt |
Closest papers by meaning
- Self-Disentanglement and Re-Composition for Cross-Domain Few-Shot Segmentation (ICML 2025 · similarity 0.91)
- Adapt Before Comparison: A New Perspective on Cross-Domain Few-Shot Segmentation (CVPR 2024 · similarity 0.91)
- The Devil is in Low-Level Features for Cross-Domain Few-Shot Segmentation (CVPR 2025 · similarity 0.90)
- Dual-Agent Optimization framework for Cross-Domain Few-Shot Segmentation (CVPR 2025 · similarity 0.90)
- Adapter Naturally Serves as Decoupler for Cross-Domain Few-Shot Semantic Segmentation (ICML 2025 · similarity 0.90)
- Cross-Domain Few-Shot Segmentation via Iterative Support-Query Correspondence Mining (CVPR 2024 · 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.
