Backtrack is a web application that helps reunite people with misplaced belongings by allowing users to report items they’ve found or lost, attach photos, and pin precise locations on a map. Powered by AI, the system analyzes image and description similarity to intelligently detect matches between lost inquiries and found reports. The platform blends a clean, modern UI with fast report flows, a location picker map, and a profile area for managing reports. It is designed to be intuitive for everyday users while remaining extensible for advanced workflows like automated similarity detection, moderation, and notifications.
At its core, the app focuses on three goals:
- Speed – Real time matching system that checks every file
- Security - Ocr_ Agent keeps track of identification labels in the item, also there is no uploading before being logged in with the Auth0 API
- Community – Forming a sense of community with support and kindness
- Connection – Centralized reporting and potential matching of lost/found items.
Some features we implemented:
- Image recognition using cosine similarity score and CLIP model.
- Gemini model that is a ocr_gate and agent
- Gemini model that is a verdict on the final descision in match.
- Auth0 API for extra security in login system
- Leaflet API for map integration and pin locations
Log in or sign up for Devpost to join the conversation.