New dataset and model improve location finding from limited ground images
Seeing Less in a Sparse World: Cross-View Geolocalization from Limited Field-of-View Observations
Computer Vision and Pattern Recognition
Summary
Finding exactly where a photo was taken can be hard when GPS doesn’t work well, like in cities with tall buildings or dense forests. The authors created a big collection of paired ground and satellite photos from around the world, using limited field-of-view images like real cameras capture. They also built a smart model that connects these different views to better guess locations. This makes it easier to locate places even with just a small part of the scene visible.
What this means in practice
- •For navigation system developers: Improve location tracking in GPS-challenged environments by using limited-view ground images matched to widespread satellite photos.
- •For disaster response teams: Enable better localization of photos taken in disaster areas where GPS signals are unreliable or unavailable.
Authors
Fahimul Aleem, Shruti Vyas
Abstract
Cross-view geolocalization aims to determine the geographic location of a ground-level image by retrieving its corresponding satellite image, offering a promising alternative for localization in Global Navigation Satellite System (GNSS)-denied or degraded environments such as urban canyons, dense forests, and disaster scenarios. Despite recent progress, existing approaches remain limited by the lack of globally diverse training datasets and benchmarks that are often geographically restricted. Furthermore, most benchmarks rely on panoramic ground views, which do not reflect the limited field-of-view imagery captured by real world cameras. To address these challenges, we introduce Cross-View World 500k (CVW500k), a large-scale globally distributed dataset containing over 500,000 ground-satellite image pairs collected across diverse geographic regions and environments. Unlike previous datasets, CVW500k uses limited field-of-view ground images, providing a more realistic and challenging setting for cross-view geolocalization. In addition, we propose GeoQueryNet, a query-based cross-view fusion transformer framework designed to bridge the viewpoint gap between ground and satellite images. The model leverages a CLIP-based visual encoder with parameter-efficient LoRA adaptation and introduces a Cross-View Alignment Module that uses learnable query tokens to extract geographically consistent features across viewpoints. Our code can be found here: https://github.com/Fahim17/Cross_View_World_Dataset.git .