FindIT: A Smart Lost and Found Management System
Losing something valuable like your keys, a wallet, or a laptop is incredibly stressful, especially in a fast-paced campus environment. Currently, most lost and found processes are fragmented, relying on pure luck, scattered social media posts, or physical notice boards with limited reach.
FindIT is designed to bridge this gap. We have built a smart, privacy-first platform that centralizes the recovery process. Instead of waiting for a chance encounter, FindIT uses structured data, AI-driven image processing, and intelligent matching to reunite people with their belongings quickly and securely.
Key Features
- AI-Powered Image Analysis: Integrates Google Gemini 2.5 Flash Vision API to automatically scan uploaded photos of “Found” items, instantly generating accurate descriptions, tags, and security questions without manual data entry.
- Advanced Fuzzy Matching: Utilizes Levenshtein distance algorithms to cross-reference lost and found reports. The system is typo-resilient, case-insensitive, and strictly enforces category isolation to surface the highest-confidence matches.
- Privacy & Gradual Disclosure: Found item details are kept under restricted visibility. The system provides secure verification mechanisms (like secret questions) that protect personal contact information until a match is confirmed by both parties.
- Automated Lifecycle Management: Features a built-in background scheduler that tracks item lifespans, issuing email warnings and automatically purging expired reports after a 7-day retention cycle to maintain database efficiency.
- Role-Based Access & Security: Secured with JWT (JSON Web Tokens) and bcrypt password hashing. Standard users are restricted to modifying only their own posts, while administrative controls ensure platform integrity.
- Responsive & Scalable UI: A mobile-first React frontend ensures seamless use across all devices, backed by a FastAPI infrastructure load-tested to handle peak campus traffic (50+ concurrent users with zero failures).
How It Works
- Report: Users submit a report. For “Found” items, users simply snap a photo, and the AI auto-fills the categorization and descriptive marks.
- Search & Match: The system continuously analyzes descriptions (color, brand, location, time) and alerts users to potential similarities despite minor typos.
- Verify: The “Gradual Disclosure” process allows the finder to verify the owner using AI-generated secret questions without revealing their identity prematurely.
- Recover: Once verified, the system safely exchanges contact info or drop-off instructions.
- Resolve & Cleanup: Recovered items are marked as claimed, and stale reports are automatically archived after the 7-day threshold.
Tech Stack
- Frontend: React, Vite, Tailwind CSS, Lucide Icons
- Backend: FastAPI, Python, SQLAlchemy
- Database: MySQL
- AI & Algorithms: Google Gemini SDK,
thefuzz(Fuzzy String Matching) - Testing: Python
unittest/pytest, Locust (Load Testing)
The Team
| Name | E-Number |
|---|---|
| Dulmina Weerasinghe | E/23/431 |
| Livindu Jayasinghe | E/23/149 |
| Lihini Silva | E/23/382 |
| Thenuk Piyathilake | E/23/274 |
Project Structure
/docs: Contains the System Requirements Specification (SRS), Testing protocols, and project documentation./backend: FastAPI source code, automated scheduled tasks, database models, and unit tests./frontend: React/Vite source code, UI components, and asset management.