I built an AI that can guess where a photo was taken
Hey Indie Hackers!
I’ve been building WhereWasTaken, an AI-powered image geolocation engine that helps people discover where photos were taken.
The idea started from a simple question:
“Can AI figure out where a photo was taken just by looking at it?”
Most images don’t have GPS coordinates or EXIF metadata. However, they still contain a lot of hidden geographic information — road signs, architecture, landscapes, vegetation, street layouts, and other visual clues.
WhereWasTaken uses AI vision and geographic reasoning to analyze these clues and estimate the most likely location.
What it can do:
📍 Find locations from photos
Upload any image and get a location prediction without GPS or metadata.
🎮 Help GeoGuessr players
Analyze screenshots and game rounds to understand geographic clues and improve deduction skills.
🔎 Investigate unknown images
Explore possible origins of images from the web or social media.
🌎 Learn geography through AI
See how visual patterns reveal differences between countries and regions.
The goal is not only to provide an answer, but also explain why the AI thinks a place matches — including evidence, confidence scores, and alternative possibilities.
I’m still improving the accuracy and user experience, and I’d love feedback from the Indie Hackers community:
Would you use an AI tool like this?
What use cases would you find most valuable?
What features would you like to see next?
Try it here:
https://wherewastaken.com
Looking forward to hearing your thoughts! 🚀
