Research map / Detection and segmentation / CNN features
Efficient Large-Scale Localization by Global Instance Recognition
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 |
|---|---|---|---|
| Use global building-instance recognition for progressive, local reference search instead of searching the whole image database | Large-scale visual localization · Aachen v1.1 | number of search frames: 6697 -> 202 (day); 6697 -> 650 (night) | Helped |
| Use progressive reference search over recognized instances | Large-scale visual localization · Aachen v1.1 | success ratio: 94.4% day and 77.0% night at the reported average search; 98.5% day and 95.8% night at top@30 | Mixed |
| Add building-instance-wise feature detection and matching (ablation) | Proposed localization method · Aachen v1.1 day | success ratio at 0.25m/2°: 86.3 -> 87.1 | Helped |
| Add building-instance-wise feature detection and matching (ablation) | Proposed localization method · Aachen v1.1 night | success ratio at 0.25m/2°: 71.7 -> 73.3 | Helped |
| Add the second-step pose refinement using additional references (ablation) | Proposed localization method · Aachen v1.1 day | success ratio at 0.25m/2°: 87.1 -> 89.1 | Helped |
| Add the second-step pose refinement using additional references (ablation) | Proposed localization method · Aachen v1.1 night | success ratio at 0.25m/2°: 73.3 -> 77.0 | Helped |
| Reduce the keypoint budget from 4k to 1k while using instance-wise detection | Proposed localization method · Aachen v1.1 night | localization accuracy at 2deg, 0.25m: 88.7 -> 88.7 | Helped |
| Use the proposed framework instead of SPP + Superglue | Aachen localization · Aachen day | success ratio at 0.25m/2°: 89.6 -> 88.3 | Hurt |
| Use the proposed framework instead of SPP + Superglue | Aachen localization · Aachen v1.1 night | success ratio at 0.5m/5°: 90.6 -> 90.1 | Hurt |
| Use the proposed framework instead of SPP + Superglue | RobotCar-Seasons localization · RobotCar-Seasons night-rain | success ratio at 5m/10°: 90.2 -> 90.0 | Hurt |
Closest papers by meaning
- From Coarse to Fine: A Matching and Alignment Framework for Unsupervised Cross-View Geo-Localization (AAAI 2025 · similarity 0.86)
- RELOCATE: A Simple Training-Free Baseline for Visual Query Localization Using Region-Based Representations (CVPR 2025 · similarity 0.85)
- Scaling Image Geo-Localization to Continent Level (NeurIPS 2025 · similarity 0.85)
- R-SCoRe: Revisiting Scene Coordinate Regression for Robust Large-Scale Visual Localization (CVPR 2025 · similarity 0.85)
- Recognition through Reasoning: Reinforcing Image Geo-localization with Large Vision-Language Models (NeurIPS 2025 · similarity 0.85)
- GRE Suite: Geo-localization Inference via Fine-Tuned Vision-Language Models and Enhanced Reasoning Chains (NeurIPS 2025 · 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.