Research map / Detection and segmentation / CNN features
R2D2: Reliable and Repeatable Detector and Descriptor
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 |
|---|---|---|---|
| add separate reliability map and AP-reliability loss (ablation) | R2D2 with repeatability only · HPatches | MMA@3: 0.639 -> 0.688 | Helped |
| add repeatability map (ablation) | R2D2 with reliability only · HPatches | MMA@3: 0.588 -> 0.688 | Helped |
| multi-scale instead of single-scale extraction (ablation) | R2D2 · HPatches | MMA@3: 0.651 -> 0.686 | Helped |
| add Aachen, style-transfer and optical-flow training pairs to web images (ablation) | R2D2 · Aachen Day-Night | localized % (5m,10deg): 77.6 -> 86.7 | Helped |
| add optical flow real pairs (F) (ablation) | R2D2 trained on W+A+S · HPatches | MMA@3: 0.667 -> 0.719 | Helped |
| larger repeatability patch size N (ablation) | R2D2 · HPatches | MMA@3 | Mixed |
| R2D2 vs D2-Net | local features · Aachen Day-Night | localized % (0.5m,2deg): 44.9 -> 45.9 | Helped |
| R2D2 vs QuadNet detector | keypoint detection · Oxford JPEG ubc (K=300) | repeatability: 0.55 -> 0.40 | Hurt |
| R2D2 vs LF-Net | local features · HPatches | M-score: 0.335 -> 0.453 | Helped |
Closest papers by meaning
- Reinforced Feature Points: Optimizing Feature Detection and Description for a High-Level Task (CVPR 2020 · similarity 0.87)
- Neural Outlier Rejection for Self-Supervised Keypoint Learning (ICLR 2020 · similarity 0.87)
- Back to 3D: Few-Shot 3D Keypoint Detection with Back-Projected 2D Features (CVPR 2024 · similarity 0.86)
- SD2Event:Self-supervised Learning of Dynamic Detectors and Contextual Descriptors for Event Cameras (CVPR 2024 · similarity 0.85)
- Towards Complementary Keypoint Detection via Mixture of Detectors (NeurIPS 2026 · similarity 0.85)
- Recurrent Feature Mining and Keypoint Mixup Padding for Category-Agnostic Pose Estimation (CVPR 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.