Research map / Image generation / Autoregressive and latent generation
You Are Here: Geolocation by Embedding Maps and Images
Noe Samano, Mengjie Zhou, Andrew Calway
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 |
|---|---|---|---|
| use learned embedded descriptors instead of fixed junction/gap binary descriptors | route localisation · StreetLearn areas not rich in junctions/gaps, ~200m routes | top-1 localisation accuracy: under 60% -> over 90% | Improved |
| add turn information to embedded descriptors (ablation) | embedded-space localisation · Wall Street | top-1 localisation accuracy | Improved |
| add turn information to embedded descriptors (ablation) | embedded-space localisation · Hudson River and Union Square, 20-location routes | top-1 localisation accuracy: over 90% | No clear effect |
| use turn information alone for route matching (ablation) | route matching · Union Square | top-1 localisation accuracy | Hurt |
| discard 50% of route candidates at each motion step (ablation) | route search · StreetLearn | top-1 localisation accuracy | No clear effect |
| use embedding dimension 16 (ablation) | embedded space · Union Square | top-k% recall | Improved |
| use small-scale S1 map tiles instead of large-scale S2 (ablation) | embedded space · Union Square | top-k% recall | Improved |
| use embedded space instead of BSD with turn information | route localisation · StreetLearn all areas | top-1 localisation accuracy | Improved |
| use embedded space on shorter routes | route localisation · StreetLearn | top-5 localisation accuracy: around 90% | Improved |
Closest papers by meaning
- GeoRanker: Distance-Aware Ranking for Worldwide Image Geolocalization (NeurIPS 2025 · similarity 0.87)
- Around the World in 80 Timesteps: A Generative Approach to Global Visual Geolocation (CVPR 2025 · similarity 0.86)
- OpenStreetView-5M: The Many Roads to Global Visual Geolocation (CVPR 2024 · similarity 0.86)
- PIGEON: Predicting Image Geolocations (CVPR 2024 · similarity 0.85)
- Vision-Language Reasoning for Geolocalization: A Reinforcement Learning Approach (AAAI 2026 · similarity 0.85)
- Recognition through Reasoning: Reinforcing Image Geo-localization with Large Vision-Language Models (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.
