GeoAxis

GeoAxis relies entirely on visual content analysis. Its AI examines architectural patterns, terrain topology, vegetation types, road markings, and signage within the image pixels. These signals are matched against a database of over 5 billion indexed reference images to triangulate a precise location — no hidden file metadata required.

The Effort slider on Global Search controls how many parallel reasoning passes the AI runs. Low effort completes in ~35 seconds with one pass, Medium in ~50 seconds with two passes, and High in ~80 seconds with four passes. Higher effort yields greater accuracy at the cost of processing time.

Standard GeoAxis uses global geo-estimation and precise geolocation modes. HyperVision™ is a city-specific Visual Place Recognition layer — trained on millions of images per target city — achieving consistent sub-10-meter accuracy even in low-light, low-resolution, or minimal-context scenarios where global models degrade significantly.

Yes. Through its HyperVision™ Enterprise tier, GeoAxis supports fully on-premise deployment. All image processing stays within the organization's own infrastructure with zero external data transmission, making it viable for law enforcement, intelligence, and regulated-industry use cases.

Uploaded full-resolution images are stored in encrypted Firebase Storage under the authenticated user's path for 30 days, allowing re-runs and re-crops. After 30 days, the original is automatically deleted via a bucket lifecycle policy. Only a 240px thumbnail persists beyond that. GeoAxis does not use uploaded images to train its AI models unless the user explicitly opts in.