tools / geospy
AI photo geolocation that predicts where an image was taken from its visual content alone (architecture, vegetation, road features) and returns candidate regions or coordinates.
field note
Treat predictions as a starting hypothesis, then verify with Street View or satellite imagery. AI geolocation can be confidently wrong, and locating private individuals from their photos raises serious privacy concerns.
used in playbooks
No-code proximity search over OpenStreetMap. Pick features you can see in an image (e.g. a park, a tram stop, a tower) and a distance, and it returns every place where they occur together.
Web IDE for querying OpenStreetMap data. Find every place matching a feature combination (e.g. a church next to a rail line within 200 m of a river) to narrow down where a photo was taken.
3D globe with high-resolution satellite imagery, 3D buildings and terrain. The desktop Pro version adds historical imagery, which you need for change-over-time analysis.
Computes the visible horizon and viewshed from any point using elevation data, and labels the peaks you'd see. Helps confirm what's visible from a claimed location.
Open, crowdsourced street-level imagery (formerly OpenStreetCam). A useful second source when Street View and Mapillary coverage is thin.
Crowd size estimator. Draw the area a crowd occupies on a map, set a density per square metre, and get a defensible estimate of attendance.