Research map / Detection and segmentation / Open-set and few-shot detection
Hierarchical Dense Correlation Distillation for Few-Shot Segmentation
Read the paperCheck 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 |
|---|---|---|---|
| use HDMNet instead of DGPNet | ResNet-50 few-shot segmenter · COCO-20i 1-shot | mIoU: 45.0 -> 50.0 | Helped |
| use HDMNet instead of DGPNet | ResNet-50 few-shot segmenter · COCO-20i 5-shot | mIoU: 56.2 -> 56.0 | Hurt |
| use HDMNet instead of BAM | ResNet-50 few-shot segmenter · PASCAL-5i 1-shot | mIoU: 67.8 -> 69.4 | Helped |
| use HDMNet instead of BAM | ResNet-50 few-shot segmenter · PASCAL-5i 5-shot | mIoU: 70.9 -> 71.8 | Helped |
| add the ensemble module (ablation) | baseline few-shot segmenter · COCO-20i 1-shot | mIoU: 44.7 -> 45.8 (+1.1) | Helped |
| decouple feature down-sampling and matching with HDM (ablation) | ensemble baseline · COCO-20i 1-shot | mIoU: 45.8 -> 47.9 (+2.1) | Helped |
| replace attention matching with correlation matching (ablation) | HDM few-shot segmenter · COCO-20i 1-shot | mIoU: 47.9 -> 48.3 (+0.4) | Helped |
| add correlation map distillation (ablation) | HDM with correlation matching · COCO-20i 1-shot | mIoU: 48.3 -> 50.0 (+1.7) | Helped |
| use cosine similarity rather than cross-attention (ablation) | HDM matching module · COCO-20i 1-shot | mIoU: 48.0 -> 49.2 | Helped |
| normalize cosine correlations with inverse softmax rather than softmax (ablation) | HDM matching module · COCO-20i 1-shot | mIoU: 49.5 -> 50.0 | Helped |
| use KL divergence with adjacent-layer distillation instead of CE with KD (ablation) | HDM training objective · COCO-20i 1-shot | mIoU: 49.1 -> 50.0 | Helped |
| increase matching stages from 1 to 3 (ablation) | HDM decoder · COCO-20i 1-shot | mIoU: 47.1 -> 50.0 | Helped |
Closest papers by meaning
- Unlocking the Potential of Pre-trained Vision Transformers for Few-Shot Semantic Segmentation through Relationship Descriptors (CVPR 2024 · similarity 0.88)
- Label-Efficient Few-Shot Semantic Segmentation with Unsupervised Meta-Training (AAAI 2024 · similarity 0.88)
- Cross-Domain Few-Shot Segmentation via Iterative Support-Query Correspondence Mining (CVPR 2024 · similarity 0.87)
- Rethinking Few-shot 3D Point Cloud Semantic Segmentation (CVPR 2024 · similarity 0.87)
- Text Augmented Correlation Transformer For Few-shot Classification & Segmentation (CVPR 2025 · similarity 0.87)
- A Surprisingly Simple Approach to Generalized Few-Shot Semantic Segmentation (NeurIPS 2024 · similarity 0.86)
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.