<div align="center">

🌏🚶‍♂️🔍CV-Cities: Advancing Cross-View Geo-Localization in Global Cities

<!-- [](https://paperswithcode.com/sota/visual-place-recognition-on-cv-cities?p=cv-cities-advancing-cross-view-geo) [](https://paperswithcode.com/sota/image-based-localization-on-cvact?p=cv-cities-advancing-cross-view-geo) [](https://paperswithcode.com/sota/image-based-localization-on-cvusa-1?p=cv-cities-advancing-cross-view-geo) [](https://paperswithcode.com/sota/image-based-localization-on-vigor-cross-area?p=cv-cities-advancing-cross-view-geo) [](https://paperswithcode.com/sota/image-based-localization-on-vigor-same-area?p=cv-cities-advancing-cross-view-geo) [](https://paperswithcode.com/sota/drone-view-target-localization-on-university-1?p=cv-cities-advancing-cross-view-geo)--> </div>

Description 📜

Cross-view geo-localization（CVGL） is beset with numerous difficulties and challenges, mainly due to the significant discrepancies in viewpoint, the intricacy of localization scenarios, and global localization needs. Given these challenges, we present a novel cross-view image geo-localization framework. The experimental results demonstrate that the proposed framework outperforms existing methods on multiple public datasets and self-built datasets. To improve the cross-view geo-localization performance of the framework on a global scale, we have built a novel global cross-view geo-localization dataset, CV-Cities. This dataset encompassing a diverse range of intricate scenarios. It serves as a challenging benchmark for cross-view geo-localization.

CV-Cities: Global Cross-view Geo-localization Dataset 💾

We collected 223,736 ground images and 223,736 satellite images with high-precision GPS coordinates of 16 typical cities in five continents. To download this dataset, you can click: 🤗CV-Cities or 🤗CV-Cities (mirror).

City distribution 📊

<td style="text-align: center"><img src="/figures/distribution_map_of_cities.png" alt="City distribution" width="600"></td>

Sample points and monthly distribution of 16 cities 📍

<table style="text-align: center"> <tr> <td style="text-align: center"><img src="/figures/captown-month.png" alt="Capetown" width="100"></td> <td style="text-align: center"><img src="/figures/london-month.png" alt="London" width="100"></td> <td style="text-align: center"><img src="/figures/melbourne-month.png" alt="Melbourne" width="100"></td> <td style="text-align: center"><img src="/figures/mexico-month.png" alt="mexico" width="100"></td> <td rowspan="8"><img src="/figures/monthly-scalebar.png" alt="Colorbar" width="40"> <br>Colorbar</td> </tr> <tr> <td style="text-align: center">Capetown, South Africa</td> <td style="text-align: center">London, UK</td> <td style="text-align: center">Melbourne, Australia</td> <td style="text-align: center">Mexico city, Mexico</td> </tr> <tr> <td style="text-align: center"><img src="/figures/newyork-month.png" alt="newyork" width="100"></td> <td style="text-align: center"><img src="/figures/paris-month.png" alt="paris" width="100"></td> <td style="text-align: center"><img src="/figures/rio-month.png" alt="rio" width="100"></td> <td style="text-align: center"><img src="/figures/taipei-month.png" alt="Taipei" width="100"></td> </tr> <tr> <td style="text-align: center">New York, USA</td> <td style="text-align: center">Paris, France</td> <td style="text-align: center">Rio, Brazil</td> <td style="text-align: center">Taipei, China</td> </tr> <tr> <td style="text-align: center"><img src="/figures/losangeles-month.png" alt="Losangeles" width="100"></td> <td style="text-align: center"><img src="/figures/maynila-month.png" alt="Maynila" width="100"></td> <td style="text-align: center"><img src="/figures/santiago-month.png" alt="Santiago" width="100"></td> <td style="text-align: center"><img src="/figures/sydney-month.png" alt="Sydney" width="100"></td> </tr> <tr> <td style="text-align: center">Losangeles, USA</td> <td style="text-align: center">Manila, Philipine</td> <td style="text-align: center">Santiago, Chile</td> <td style="text-align: center">Sydney, Australia</td> </tr> <tr> <td style="text-align: center"><img src="/figures/seattle-month.png" alt="Seattle" width="100"></td> <td style="text-align: center"><img src="/figures/singapore-month.png" alt="Singapore" width="100"></td> <td style="text-align: center"><img src="/figures/barcelona-month.png" alt="Barcelona" width="100"></td> <td style="text-align: center"><img src="/figures/tokyo-month.png" alt="Tokyo" width="100"></td> </tr> <tr> <td style="text-align: center">Seattle, USA</td> <td style="text-align: center">Singapore</td> <td style="text-align: center">Barcelona, Span</td> <td style="text-align: center">Tokyo, Japan</td> </tr> </table>

Different scenes 🏞️

<table> <tr> <td style="text-align: center"><img src="/figures/figure2-1-1.jpg" alt="ground image" width="150"></td> <td style="text-align: center"><img src="/figures/figure2-1-2.jpg" alt="satellite image" width="75"></td> <td style="text-align: center"><img src="/figures/figure2-1-3.jpg" alt="ground image" width="150"></td> <td style="text-align: center"><img src="/figures/figure2-1-4.jpg" alt="satellite image" width="75"></td> </tr> <tr> <td style="text-align: center" colspan="2">City scene</td> <td style="text-align: center" colspan="2">Nature scene</td> </tr> <tr> <td style="text-align: center"><img src="/figures/figure2-2-1.jpg" alt="ground image" width="150"></td> <td style="text-align: center"><img src="/figures/figure2-2-2.jpg" alt="satellite image" width="75"></td> <td style="text-align: center"><img src="/figures/figure2-2-3.jpg" alt="ground image" width="150"></td> <td style="text-align: center"><img src="/figures/figure2-2-4.jpg" alt="satellite image" width="75"></td> </tr> <tr> <td style="text-align: center" colspan="2">Water scene</td> <td style="text-align: center" colspan="2">Occlusion</td> </tr> </table>

Scenes, yearly and monthly distribution 📊

<table> <tr> <td style="text-align: center"><img src="/figures/figure3a.png" alt="Scenes distribution" width="200"></td> <td style="text-align: center"><img src="/figures/figure3b.png" alt="Yearly distribution" width="200"></td> <td style="text-align: center"><img src="/figures/figure3c.png" alt="monthly distribution" width="200"></td> </tr> </table>

Framework 🖇️

<td style="text-align: center"><img src="/figures/figure4.png" alt="Framework" width="500"></td>

Precision distribution 🚿

<table style="text-align: center"> <tr> <td style="text-align: center"><img src="/figures/precision_london100.jpg" alt="London" width="150"></td> <td style="text-align: center"><img src="/figures/precision_rio100.jpg" alt="Rio" width="150"></td> <td style="text-align: center"><img src="/figures/precision_seattle100.jpg" alt="seattle" width="150"></td> </tr> <tr> <td style="text-align: center">London, UK</td> <td style="text-align: center">Rio, Brazil</td> <td style="text-align: center">Seattle, USA</td> </tr> <tr> <td style="text-align: center"><img src="/figures/precision_sinapore100.jpg" alt="Singapore" width="150"></td> <td style="text-align: center"><img src="/figures/precision_sydney100.jpg" alt="sydney" width="150"></td> <td style="text-align: center"><img src="/figures/precision_taipei100.jpg" alt="taipei" width="150"></td> </tr> <tr> <td style="text-align: center">Singapore</td> <td style="text-align: center">Sydney, Australia</td> <td style="text-align: center">Taipei, China</td> </tr> </table>

Model Zoo 📦

🚧 Under Construction

Train the CVCities 🚂

python train/train_cvcities.py

Acknowledgments 🧭

This code is based on the amazing work of:

• DINOv2
• Sample4Geo
• QGIS

Citation✅

  @ARTICLE{huangCVCities2024,
  author={Huang, Gaoshuang and Zhou, Yang and Zhao, Luying and Gan, Wenjian},
  journal={IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing}, 
  title={CV-Cities: Advancing Cross-View Geo-Localization in Global Cities}, 
  year={2025},
  volume={18},
  number={},
  page