Research map / Label-efficient and robust learning / Recognition and long-tail
Local Descriptors Optimized for Average Precision
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 |
|---|---|---|---|
| optimize matching with listwise Average Precision rather than triplet-based losses | DOAP descriptors · UBC Phototour | FPR95 mean: 1.90 -> 1.45 | Helped |
| add a Spatial Transformer (ablation) | DOAP real-valued descriptors · UBC Phototour | FPR95 mean: 1.45 -> 1.38 | Helped |
| add a Spatial Transformer (ablation) | Binary DOAP · UBC Phototour | FPR95 mean: 3.01 -> 2.89 | Helped |
| add a Spatial Transformer (ablation) | DOAP · RomePatches | mAP | No clear effect |
| add clustering-based label mining for in-sequence distractors (ablation) | DOAP-ST · HPatches image matching | mAP: around 6% | Helped |
| add clustering-based label mining for in-sequence distractors (ablation) | DOAP-ST · HPatches patch retrieval | mAP | Helped |
| sample more patches from the same image sequences when label mining is used (ablation) | DOAP-ST-LM · HPatches image matching | mAP: about 6% absolute | Helped |
| increase training batch size (ablation) | DOAP · UBC Phototour | FPR95 | Helped |
| use default zero padding instead of boundary replication padding (ablation) | DOAP with Spatial Transformer · HPatches | descriptor matching performance | No clear effect |
Closest papers by meaning
- Comparative Evaluation of Hand-Crafted and Learned Local Features (CVPR 2017 · similarity 0.89)
- HPatches: A Benchmark and Evaluation of Handcrafted and Learned Local Descriptors (CVPR 2017 · similarity 0.89)
- Local Features and Visual Words Emerge in Activations (CVPR 2019 · similarity 0.87)
- FeatureBooster: Boosting Feature Descriptors With a Lightweight Neural Network (CVPR 2023 · similarity 0.87)
- Kernelized Subspace Pooling for Deep Local Descriptors (CVPR 2018 · similarity 0.87)
- Focus on Local: Finding Reliable Discriminative Regions for Visual Place Recognition (AAAI 2025 · similarity 0.84)
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.