GeoSolvr

Project Overview

GeoSolvr mimics the logical process GeoGuessr pros use: the image gets split up and handled by different agents, then brought back together and improved in iterations.
The goal is a synchronous orchestration which provides analysis on every single output in hopes of teaching you the reasoning behind country identification on Street View!

Features

• Uses in depth reasoning to predict location
• Runs multiple times and compares responses against eachother for accuracy
• Uses advanced OCR tool to detect and analyze text in images
• Uses 4 seperate agents that communicate to reach a final decision

What I learned

• This was my first time building a Multi Agent AI system from scratch
• I learned about implemeting different tools to improve accuaracy and speed
• Learned about the ethical peice of the system
• Got to analyze real GeoGuessr pros and use my knowledge to help me build my system

View the Code

The full source code is available on GitHub.