Daniel Heinen
@heinenbros
Serve others before yourself. Founder of @graylark_io and @GeospyAI
🚨 Launching Today: GeoSpy Street Targeting – Greater London 🛰️ 600+ square miles. 3.6M buildings. 60K streets. GeoSpy now delivers meter-level geolocation from pixels alone — no metadata needed. #GeoSpy #OSINT #GeoAI #London
People think LLMs are smart enough to geolocate any image. So I built an AI that poisons images. Give it a photo, and it spoofs the location using pixels only ChatGPT and Gemini Fall for it GeoSpy does not Drop a comment if I should launch this as an app
If building @GeospyAI has taught me one thing. Taking AI research and turning it into a product is both extremely difficult and extremely rewarding. You can build things people call magical. Every time we do a demo with real life cases… peoples jaws drop.
🚁🔍 GeoSpy’s new vision model can take a shaky, low-altitude drone clip—no GPS, no Remote ID—and rebuild the whole flight path. Pixels in, coordinates out. Lost UAVs, contraband drops, no-fly incursions: tracked. #GeoSpy #CounterUAS #DroneForensics #AI
Mind blowing that GeoSpy 2.0 knows what San Fransisco Sidewalk looks like. AI can now locate a photo of your shoes. Curios how well this works in other cities.
After meeting everyone at @interaction I immediately hired a German CTO. Best decision of my life
This will be an absolute game changer for investigators working on combatting ICAC (Internet Crimes Against Children). Great work @heinenbros
We trained an AI model on hundreds of millions of images: It can take an ultra low-context indoor photo—no windows, no signs—and still predict an exact address.
This is what happens when you train an AI on high-quality building image data. People didn’t believe my last post about GeoSpy locating indoor scenes— So here’s a more realistic, ultra low-context example.
Yea man there’s no way this is possible lol
Yea man there’s no way this is possible lol
GeoSpy is able to track the location of a photo taken indoors.