Research map / Label-efficient and robust learning / Domain adaptation and generalization
Remember the Difference: Cross-Domain Few-Shot Semantic Segmentation via Meta-Memory Transfer
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 |
|---|---|---|---|
| transfer source-domain meta-memory styles to target features | CDFSS · COCO-20i to PASCAL-5i, 1-shot | mean-IoU: 57.4 -> 65.6 | Helped |
| use CDFSS instead of HSNet | cross-domain few-shot segmentation · COCO-20i to PASCAL-5i, ResNet50, 1-shot | mIoU: 61.5 -> 65.6 | Helped |
| use CDFSS instead of HSNet | cross-domain few-shot segmentation · COCO-20i to FSS-1000 | mIoU: 80.8 -> 81.9 | Helped |
| use CDFSS instead of HSNet | cross-domain few-shot segmentation · COCO-20i to SUIM | mIoU: 35.1 -> 37.1 | Helped |
| use CDFSS instead of SCL | cross-domain few-shot segmentation · PASCAL-5i to SUIM | mIoU: 31.8 -> 34.7 | Helped |
| add contrastive loss (ablation) | ASGNet · COCO-20i to PASCAL-5i, 1-shot | mean-IoU: 57.4 -> 62.8 | Helped |
| add memory-enhanced source features to contrastive loss (ablation) | ASGNet · COCO-20i to PASCAL-5i, 1-shot | mean-IoU: 62.8 -> 63.5 | Helped |
| add memory-enhanced target features to the contrastive-loss model with MEnS (ablation) | ASGNet · COCO-20i to PASCAL-5i, 1-shot | mean-IoU: 63.5 -> 65.6 | Helped |
| replace meta-memory with random noise having the same mean and variance (ablation) | CDFSS · COCO-20i to PASCAL-5i | performance: 6.1% | Hurt |
| place the memory module at layer-0 rather than using no memory (ablation) | ResNet50 segmentation model · COCO-20i to PASCAL-5i, 1-shot | mean-IoU: 62.8 -> 64.0 | Helped |
| place the memory module at layer-1 rather than using no memory (ablation) | ResNet50 segmentation model · COCO-20i to PASCAL-5i, 1-shot | mean-IoU: 62.8 -> 62.7 | No clear effect |
| place the memory module at layer-2 rather than using no memory (ablation) | ResNet50 segmentation model · COCO-20i to PASCAL-5i, 1-shot | mean-IoU: 62.8 -> 53.3 | Hurt |
Closest papers by meaning
- The Devil is in Low-Level Features for Cross-Domain Few-Shot Segmentation (CVPR 2025 · similarity 0.91)
- Unlocking the Potential of Pre-trained Vision Transformers for Few-Shot Semantic Segmentation through Relationship Descriptors (CVPR 2024 · similarity 0.90)
- Adapter Naturally Serves as Decoupler for Cross-Domain Few-Shot Semantic Segmentation (ICML 2025 · similarity 0.90)
- Label-Efficient Few-Shot Semantic Segmentation with Unsupervised Meta-Training (AAAI 2024 · similarity 0.90)
- Cross-Domain Few-Shot Segmentation via Iterative Support-Query Correspondence Mining (CVPR 2024 · similarity 0.89)
- Adapt Before Comparison: A New Perspective on Cross-Domain Few-Shot Segmentation (CVPR 2024 · 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.