Research map / Detection and segmentation / Open-set and few-shot detection
Doubly Deformable Aggregation of Covariance Matrices for Few-Shot Segmentation
Zhitong Xiong, Haopeng Li, Xiao Xiang Zhu
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 GP-based kernel learning (GP-KL) (ablation) | HSNet baseline · PASCAL-5i 1-shot, VGG16 | mIoU: 59.6 -> 60.8 | Improved |
| add GP-based kernel learning (GP-KL) (ablation) | HSNet baseline · PASCAL-5i 5-shot, VGG16 | mIoU: 64.9 -> 65.2 | Improved |
| add one DDT layer after GP-KL (ablation) | HSNet + GP-KL · PASCAL-5i 1-shot, VGG16 | mIoU: 60.8 -> 61.2 | Improved |
| use two DDT layers instead of one (ablation) | HSNet + GP-KL + DDT · PASCAL-5i 1-shot, VGG16 | mIoU: 61.2 -> 61.8 | Improved |
| use three DDT layers instead of two (ablation) | HSNet + GP-KL + DDT · PASCAL-5i 1-shot, VGG16 | mIoU: 61.8 -> 61.6 | Hurt |
| use three DDT layers instead of two (ablation) | HSNet + GP-KL + DDT · PASCAL-5i 5-shot, VGG16 | mIoU: 66.1 -> 65.8 | Hurt |
| combine GP-KL and DDT (ablation) | HSNet · PASCAL-5i 1-shot, VGG16 | mIoU: 59.6 -> 61.8 | Improved |
| plug DACM covariance and DDT into VAT | VAT · PASCAL-5i 1-shot, ResNet50 | mIoU: 65.3 -> 66.8 | Improved |
| plug DACM covariance and DDT into VAT | VAT · COCO-20i 1-shot, ResNet50 | mIoU: 41.3 -> 43.0 | Improved |
| plug DACM covariance and DDT into VAT | VAT · PASCAL-5i 5-shot, ResNet101 | mIoU: 71.6 -> 73.3 | Improved |
| DACM alone vs VAT | VAT · PASCAL-5i 1-shot, ResNet101 | mIoU: 67.5 -> 67.5 | No clear effect |
| DACM alone vs VAT | VAT · PASCAL-5i 5-shot, ResNet101 | mIoU: 71.6 -> 71.4 | Hurt |
Closest papers by meaning
- Cost Aggregation with 4D Convolutional Swin Transformer for Few-Shot Segmentation (ECCV 2022 · similarity 0.87)
- Hierarchical Dense Correlation Distillation for Few-Shot Segmentation (CVPR 2023 · similarity 0.87)
- Unlocking the Potential of Pre-trained Vision Transformers for Few-Shot Semantic Segmentation through Relationship Descriptors (CVPR 2024 · similarity 0.86)
- Label-Efficient Few-Shot Semantic Segmentation with Unsupervised Meta-Training (AAAI 2024 · similarity 0.86)
- Cross-Domain Few-Shot Semantic Segmentation (ECCV 2022 · similarity 0.86)
- Rethinking Few-shot 3D Point Cloud Semantic Segmentation (CVPR 2024 · similarity 0.85)
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.
