Research map / Learning theory and algorithms / Prediction and decision losses
Robust and Decomposable Average Precision for Image Retrieval
Elias Ramzi, Nicolas THOME, Clément Rambour, Nicolas Audebert, Xavier Bitot
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 |
|---|---|---|---|
| Replace SmoothAP with ROADMAP | ResNet-50 image retrieval · CUB-200-2011 | mAP@R: 23.9 -> 25.3 | Improved |
| Replace SmoothAP with ROADMAP | ResNet-50 image retrieval · SOP | mAP@R: 54.6 -> 56.5 | Improved |
| Replace SmoothAP with ROADMAP | ResNet-50 image retrieval · INaturalist-2018 | mAP@R: 20.7 -> 25.1 | Improved |
| Replace XBM with ROADMAP | ResNet-50 image retrieval · SOP | R@1: 80.6 -> 82.0 | Improved |
| Replace XBM with ROADMAP | ResNet-50 image retrieval · INaturalist-2018 | mAP@R: 18.5 -> 25.1 | Improved |
| Add the H− robust rank approximation, using SupAP instead of SmoothAP (ablation) | ResNet-50 image retrieval · CUB-200-2011 | mAP@R: 23.9 -> 24.6 | Improved |
| Add the H− robust rank approximation, using SupAP instead of SmoothAP (ablation) | ResNet-50 image retrieval · SOP | mAP@R: 54.6 -> 55.3 | Improved |
| Add the H− robust rank approximation, using SupAP instead of SmoothAP (ablation) | ResNet-50 image retrieval · INaturalist-2018 | mAP@R: 20.7 -> 21.3 | Improved |
| Add calibration loss to SupAP (ablation) | ResNet-50 image retrieval · CUB-200-2011 | mAP@R: 24.6 -> 25.3 | Improved |
| Add calibration loss to SupAP (ablation) | ResNet-50 image retrieval · SOP | mAP@R: 55.3 -> 56.5 | Improved |
| Add calibration loss to SupAP (ablation) | ResNet-50 image retrieval · INaturalist-2018 | mAP@R: 21.3 -> 25.1 | Improved |
| Add calibration loss to SupAP (ablation) | ResNet-50 image retrieval with batch size 32 · INaturalist-2018 | mAP@R: 17.13 -> 20.43 | Improved |
Closest papers by meaning
- Optimization of Rank Losses for Image Retrieval (TPAMI 2025 · similarity 0.92)
- Learning With Average Precision: Training Image Retrieval With a Listwise Loss (ICCV 2019 · similarity 0.90)
- Smooth-AP: Smoothing the Path Towards Large-Scale Image Retrieval (ECCV 2020 · similarity 0.89)
- Hierarchical Average Precision Training for Pertinent Image Retrieval (ECCV 2022 · similarity 0.89)
- Threshold-Consistent Margin Loss for Open-World Deep Metric Learning (ICLR 2024 · similarity 0.87)
- Collapse-Aware Triplet Decoupling for Adversarially Robust Image Retrieval (ICML 2024 · 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.
