Post by Casey Ashe

🚀 I've just open-sourced SightPoint AI — a desktop OSINT geolocation tool I built from scratch. Upload an image. Get coordinates. SightPoint AI uses Google Gemini 2.5 Flash Vision to perform forensic analysis of images and estimate where in the world they were taken — entirely within a self-contained desktop interface. No browser. No local server. No external windows. What it analyses: → Architecture style, road infrastructure, vegetation, vehicle types → Licence plate formats and country identification → Railway gauge estimation from sleeper spacing → Aviation tail number prefix detection → Language and script identification → Sun azimuth and shadow analysis for hemisphere estimation → Full OCR of all visible text → Embedded EXIF GPS metadata Results are rendered on an interactive embedded map (road, satellite, hybrid, terrain) with a confidence score, heatmap of candidate locations, and a full chain-of-evidence reasoning breakdown. Every analysis is logged to a local CSV history file. Built entirely in Python using tkinter, PIL, Claude Sonnet and the Gemini Vision API. This is part of my broader work building an OSINT tool portfolio as I work toward a career in digital forensics and intelligence. I attended Infosec Europe earlier this year and this project reflects what I've been building since — practical, ethical, analyst-focused tooling. 🔗 GitHub: https://lnkd.in/ec-jtSU9 Feedback from practitioners in the OSINT, forensics, and intelligence community is very welcome. #OSINT #DigitalForensics #Cybersecurity #Python #OpenSource #Geolocation #IntelligenceAnalysis