This platform finds the location of any photo using AI
Geospy.ai — Screenshot © Geospy AI
An online platform aims to allow Internet users to find the location from which a photo was taken. To do this, the technology uses artificial intelligence and in particular computer vision to analyze different elements and search for the location represented.
Find the location of a photo
Geospy allows Internet users to find the location where a photograph was taken in a few clicks. Belonging to the OSINT (Open Source Intelligence) family of tools, this technology could benefit government authorities in legal investigations. In a broader scope, Geospy is also aimed at journalists and the insurance sector.
Concretely, when the user uploads a photo to the platform, it is analyzed by an AI which takes into account several elements, including the architecture of the place or any store fronts. While the company behind the project says accuracy is much higher in San Francisco than in other cities at this point, some Geospy models can be tried for all locations on a global scale.
Our curiosity piqued, we tried using Geospy. For this, we transferred a photograph taken on the Place des Arcades in Valbonne, a village located in the Alpes-Maritimes.
Test location with Geospy.ai — Screenshot © Florent Lanne for Les Numériques
If Geospy does not correctly locate the location of the photo, it still manages to locate within a radius of 30 km as the crow flies. Here, the platform indicates that the photo was taken in Vence, a village whose architecture is significantly close to that of the Place des Arcades in Valbonne.
The road ahead remains long
In its current version, Geospy sometimes presents some inconsistencies. We sent a second photo in which a parked vehicle appears in the foreground. The photo was taken on the outskirts of Bordeaux.
Geospy.ai Trial — Screenshot © Florent Lanne for Les Numériques
In this second attempt, the platform makes a mistake and indicates that the photo was taken in Toulouse. What's interesting is how Geospy reasoned during the analysis.
The white car's license plate starts with “GS”, which is a regional identifier for Haute-Garonne, a department in southwestern France. The architecture of the buildings in the background, with their tiled roofs and shutters, is typical of French residential areas. The vegetation, including trees and shrubs, is consistent with the flora found in the south of France. Researching street views in Toulouse, France reveals similar architecture and vegetation.
In addition to the error linked to the regional identifier of the plate – although registered in 33, which corresponds to Gironde – the reasoning of the tool confirms that it seeks to rely on elements present in the photos to issue logical deductions.
