Breaking

Netryx Astra V2 pinpoints location from a single image

July 7, 2026 at 01:08 EDT

Open Source
AI Agents
Research & Papers

An open-source geolocation tool called Netryx Astra V2, which can estimate GPS coordinates to within a few meters from a single image, has been released. Working from pixel data alone—without GPS or metadata—it identifies where a photo was taken by matching it against a street-view image database, and is available for free on GitHub. The tool was built by a student developer (GitHub account sparkyniner), and the V2 release was introduced on Reddit's r/OSINT.

Open-Source · Netryx Astra V2

One Photo In, GPS Coordinates Out — Accurate to a Few Meters

A newly released, free tool geolocates a single street-level image using pure computer vision — no metadata, no large language models — matching it against a database of Street View–style panoramas.

≤ few m

Location accuracy from a single image

3 stages

Pipeline, down from 9+ in V1

~min

Search time per query, runs locally

Pipeline: from 9+ stages to 3

V2 simplifies the process while improving accuracy. Column height = number of stages.

V1 stages

V2 stages

The three-stage matching flow

MegaLoc

Image retrieval narrows down candidate locations

→

MASt3R

Dense 3D matching verifies pixel-level correspondence

→

Spatial Consensus

Geographic clustering filters out false matches

Index build time grows with coverage radius

Users build a local database from Street View imagery — larger areas take far longer.

1 km

20–30 min

5 km

a few hours

10 km

~overnight

What it can do

Works with no metadata at all
Handles blurry, cropped, zoomed or flipped images
Even locates from window-reflected images
Free, fully open source, runs locally offline

Limitations & concerns

Only works where the index has coverage
High compute cost — not for real-time use
Can freeze if memory runs low
Raises privacy and misuse questions

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.