Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Few-Shot Learning With Localization in Realistic Settings
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 |
|---|---|---|---|
| Add batch folding | Prototypical Networks · meta-iNat | Top-1 mean accuracy: 16.07$\pm$.19 -> 20.04$\pm$.04 | Helped |
| Add few-shot localization to batch-folded Prototypical Networks | PN+BF · meta-iNat | Top-1 mean accuracy: 20.04$\pm$.04 -> 26.25$\pm$.05 | Helped |
| Use unsupervised instead of few-shot localization with batch folding | PN+BF · meta-iNat | Top-1 mean accuracy: 20.04$\pm$.04 -> 28.75$\pm$.13 | Helped |
| Add covariance pooling to batch folding and few-shot localization | PN+BF+fsL · meta-iNat | Top-1 mean accuracy: 26.25$\pm$.05 -> 35.52$\pm$.05 | Helped |
| Add covariance pooling to batch folding and unsupervised localization | PN+BF+usL · meta-iNat | Top-1 mean accuracy: 28.75$\pm$.13 -> 32.74$\pm$.13 | Helped |
| Add covariance pooling (ablation) | Prototypical Networks · meta-iNat | Top-1 mean accuracy: 16.07$\pm$.19 -> $24.31\pm.48$ | Helped |
| Add localization without batch folding or covariance pooling (ablation) | Prototypical Networks · meta-iNat | Top-1 mean accuracy: 16.07$\pm$.19 -> $21.55\pm.09$ | Helped |
| Add batch folding on mini-ImageNet | Prototypical Networks · mini-ImageNet, 5-shot | Accuracy: 65.76$\pm$.29 -> 65.2$\pm$.29 | Hurt |
| Add few-shot localization | Prototypical Networks · mini-ImageNet, 5-shot | Accuracy: 65.76$\pm$.29 -> 67.85$\pm$.29 | Helped |
| Add few-shot localization and covariance pooling | Prototypical Networks · mini-ImageNet, 1-shot | Accuracy: 49.97$\pm$.30 -> 49.64$\pm$.31 | No clear effect |
| Use covariance pooling | Localized model · meta-iNat, class-size comparison | Top-1 accuracy | Mixed |
| Train the few-shot localizer jointly with the classifier rather than use an untrained localizer (ablation) | Few-shot localization model with batch folding · meta-iNat | Top-1 mean accuracy: 19.74$\pm$.03 -> 26.28$\pm$.05 | Helped |
Closest papers by meaning
- Meta-Learning With Differentiable Convex Optimization (CVPR 2019 · similarity 0.90)
- Pushing the Limit of Fine-Tuning for Few-Shot Learning: Where Feature Reusing Meets Cross-Scale Attention (AAAI 2024 · similarity 0.89)
- Pushing the Limits of Simple Pipelines for Few-Shot Learning: External Data and Fine-Tuning Make a Difference (CVPR 2022 · similarity 0.89)
- A Closer Look at Few-shot Classification Again (ICML 2023 · similarity 0.88)
- Dense Classification and Implanting for Few-Shot Learning (CVPR 2019 · similarity 0.88)
- Making Large Vision Language Models to Be Good Few-Shot Learners (AAAI 2025 · similarity 0.87)
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.