Research map / 3D vision / Depth and stereo
GeoFlow: Real-Time Fine-Grained Cross-View Geolocalization via Iterative Flow Prediction
Ayesh Abu Lehyeh, Xiaohan Zhang, Ahmad Arrabi, Waqas Sultani, Chen Chen, Safwan Wshah
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 full IRS (N=10, R=5) instead of single pass (N=1, R=1) (ablation) | GeoFlow · KITTI cross-area | mean error (m): 12.47 -> 8.42 | Improved |
| use full IRS instead of single pass (ablation) | GeoFlow · KITTI cross-area | median error (m): 11.79 -> 5.60 | Improved |
| increase refinement rounds from 1 to 3 (ablation) | GeoFlow (N=10) · KITTI cross-area | mean error (m): 10.69 -> 8.47 | Improved |
| increase refinement rounds from 5 to 10 (ablation) | GeoFlow (N=10) · KITTI cross-area | mean error (m) and FPS: 8.42 -> 8.41 m; 29.49 -> 26.23 FPS | No clear effect |
| increase initial seeds from 1 to 20 (ablation) | GeoFlow (R=5) · KITTI cross-area | mean error (m): 8.58 -> 8.41 | Improved |
| increase initial seeds from 10 to 20 (ablation) | GeoFlow (R=5) · KITTI cross-area | mean error (m) and FPS: 8.42 -> 8.41 m; 29.49 -> 28.08 FPS | No clear effect |
| use GeoFlow instead of CCVPE | FG-CVG · KITTI cross-area | mean error (m): 9.16 -> 8.42 | Improved |
| use GeoFlow instead of HC-Net / FG2 on accuracy | FG-CVG · VIGOR cross-area | mean error (m): HC-Net 3.35, FG2 2.74 vs GeoFlow 4.62 | Hurt |
| use GeoFlow instead of HC-Net on speed and size | FG-CVG · KITTI | FPS / memory (MiB): 25.00 -> 29.49 FPS; 1900 -> 686 MiB | Improved |
Closest papers by meaning
- Around the World in 80 Timesteps: A Generative Approach to Global Visual Geolocation (CVPR 2025 · similarity 0.89)
- GeoRouter: Dynamic Paradigm Routing for Worldwide Image Geolocalization (NeurIPS 2026 · similarity 0.89)
- GeoCLIP: Clip-Inspired Alignment between Locations and Images for Effective Worldwide Geo-localization (NeurIPS 2023 · similarity 0.88)
- Towards Interactive Global Geolocation Assistant (ECCV 2026 · similarity 0.88)
- GeoRanker: Distance-Aware Ranking for Worldwide Image Geolocalization (NeurIPS 2025 · similarity 0.88)
- Vision-Language Reasoning for Geolocalization: A Reinforcement Learning Approach (AAAI 2026 · similarity 0.87)
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.
