Statewide Visual Geolocalization in the Wild

TL;DR

This paper introduces a scalable visual geolocalization method that accurately predicts the location of street-view images within a state-sized region by matching them to aerial imagery in a joint embedding space.

Contribution

It proposes a novel layout for large-scale geographical search regions and a model that leverages multi-level aerial imagery for precise geolocation in the wild.

Findings

Successfully localizes 60.6% of street-view photos within 50m in Massachusetts.

Uses a joint embedding space for matching street-view and aerial images.

Scales to large geographical regions with consistent cell resolutions.

Abstract

This work presents a method that is able to predict the geolocation of a street-view photo taken in the wild within a state-sized search region by matching against a database of aerial reference imagery. We partition the search region into geographical cells and train a model to map cells and corresponding photos into a joint embedding space that is used to perform retrieval at test time. The model utilizes aerial images for each cell at multiple levels-of-detail to provide sufficient information about the surrounding scene. We propose a novel layout of the search region with consistent cell resolutions that allows scaling to large geographical regions. Experiments demonstrate that the method successfully localizes 60.6% of all non-panoramic street-view photos uploaded to the crowd-sourcing platform Mapillary in the state of Massachusetts to within 50m of their ground-truth location. Source code is available at https://github.com/fferflo/statewide-visual-geolocalization.