🔍 FindSafe.ai

AI-Powered Missing Person Detection System

FindSafe.ai is a full-stack AI system that helps locate missing persons using face recognition.
The public uploads sightings → AI matches faces → Family gets instant email alerts with location.

---

📸 Screenshots

---

✨ Features

• 🧠 AI Face Recognition — DeepFace (VGG-Face model) with cosine similarity matching
• 📸 Public Sighting Portal — Anyone can upload a photo, no login required
• 📍 Auto Location Capture — GPS coordinates captured on every sighting
• 🚨 Instant Email Alerts — Family notified with match confidence %, Google Maps link, and sighting photo
• 👮 Multi-Organization Auth — Police stations and NGOs register and get approved by Super Admin
• 🔄 Rescan Feature — Police can upload new photo and rescan all existing sightings
• 🛡️ Rate Limiting — Max 5 uploads per IP per hour to prevent abuse
• 🗑️ Case Management — Add, view, expand, and delete missing person cases
• 👑 Super Admin Portal — Hidden portal to approve/reject organization registrations

---

🏗️ System Architecture

┌─────────────────────────────────────────────────────────┐
│                     3 Portals                           │
├─────────────────┬──────────────────┬────────────────────┤
│  Public Portal  │   Admin Portal   │  Super Admin       │
│  (No login)     │  (Police / NGO)  │  (System Owner)    │
│                 │                  │                    │
│  Upload photo   │  Add missing     │  Approve/reject    │
│  Auto location  │  person          │  organizations     │
│  Anonymous      │  View cases      │                    │
│                 │  Rescan          │                    │
└────────┬────────┴────────┬─────────┴────────────────────┘
         │                 │
         ▼                 ▼
┌─────────────────────────────────┐
│         Flask REST API          │
│  Rate limiting · JWT-style auth │
│  File validation · CORS         │
└───────────────┬─────────────────┘
                │
        ┌───────┴────────┐
        ▼                ▼
┌──────────────┐  ┌─────────────────────────────┐
│  MySQL DB    │  │   DeepFace AI Engine         │
│              │  │                              │
│  sightings   │  │  VGG-Face model              │
│  persons     │  │  Cosine similarity matching  │
│  orgs        │  │  Face encoding storage       │
│  matches     │  │                              │
└──────────────┘  └──────────────┬───────────────┘
                                 │
                                 ▼
                  ┌──────────────────────────────┐
                  │      Gmail SMTP Alert         │
                  │                              │
                  │  Match confidence %           │
                  │  Google Maps location link   │
                  │  Sighting photo attached     │
                  └──────────────────────────────┘

---

🛠️ Tech Stack

Layer	Technology
Frontend	React.js 18, Axios, React Router
Backend	Python, Flask, Flask-CORS, Flask-Limiter
AI / ML	DeepFace, VGG-Face Model, OpenCV
Database	MySQL, python-dotenv
Email	Gmail SMTP, Python smtplib
Auth	Token-based, Werkzeug password hashing
Storage	Local file system (AWS S3 ready)

---

🚀 Getting Started

Prerequisites

• Python 3.10+
• Node.js 18+
• MySQL 8.0+
• Gmail account with App Password enabled

1. Clone the repositories

# Backend
git clone https://github.com/user1-prajwal/findsafe-backend.git
cd findsafe-backend

# Frontend (separate repo)
git clone https://github.com/user1-prajwal/findsafe-frontend.git

2. Backend Setup

# Install dependencies
pip install -r requirements.txt

# Create environment file
cp .env.example .env
# Edit .env with your MySQL and Gmail credentials

3. Database Setup

CREATE DATABASE missing_finder;
USE missing_finder;

CREATE TABLE organizations (
    id          INT AUTO_INCREMENT PRIMARY KEY,
    name        VARCHAR(100) NOT NULL,
    type        ENUM('police', 'ngo', 'superadmin') DEFAULT 'ngo',
    email       VARCHAR(100) UNIQUE NOT NULL,
    password    VARCHAR(255) NOT NULL,
    is_approved BOOLEAN DEFAULT FALSE,
    created_at  TIMESTAMP DEFAULT CURRENT_TIMESTAMP
);

CREATE TABLE missing_persons (
    id           INT AUTO_INCREMENT PRIMARY KEY,
    name         VARCHAR(100),
    age          INT,
    image_path   VARCHAR(255),
    encoding     LONGTEXT,
    family_email VARCHAR(100),
    added_by     INT,
    added_at     TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
    FOREIGN KEY (added_by) REFERENCES organizations(id)
);

CREATE TABLE public_sightings (
    id          INT AUTO_INCREMENT PRIMARY KEY,
    image_path  VARCHAR(255),
    encoding    LONGTEXT,
    latitude    DECIMAL(10, 8),
    longitude   DECIMAL(11, 8),
    uploaded_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
);

Then create the Super Admin account:

python fix_password.py

4. Run Backend

python app.py
# Runs on http://localhost:5000

5. Frontend Setup

npm install
npm start
# Runs on http://localhost:3000

---

📡 API Endpoints

Public

Method	Endpoint	Description
POST	/api/public/upload	Upload sighting photo + location
GET	/api/stats	Get real-time system stats

Auth

Method	Endpoint	Description
POST	/api/auth/register	Register new organization
POST	/api/auth/login	Login (all org types)

Admin (requires token)

Method	Endpoint	Description
POST	/api/admin/add-missing	Add missing person + scan
GET	/api/admin/missing-persons	Get all cases
POST	/api/admin/rescan	Rescan with new photo
DELETE	/api/admin/delete-person/:id	Delete a case

Super Admin (requires super token)

Method	Endpoint	Description
GET	/api/superadmin/pending	View pending registrations
GET	/api/superadmin/approved	View approved organizations
POST	/api/superadmin/approve/:id	Approve an organization
DELETE	/api/superadmin/reject/:id	Reject an organization

---

🔒 Security Features

• Rate Limiting — 5 uploads/hour per IP (Flask-Limiter)
• File Validation — Type check (JPG/PNG only) + 5MB size limit
• Password Hashing — Werkzeug bcrypt hashing
• Token Auth — Role-based access on all admin routes
• Safe Updates — No passwords or encodings exposed to frontend
• CORS — Configured for frontend origin only

---

📁 Project Structure

MPF-backend/
├── app.py              # Flask API — all routes
├── database.py         # MySQL connection + CRUD operations
├── matcher.py          # Face matching engine (cosine similarity)
├── email_service.py    # Gmail SMTP alert with HTML email
├── fix_password.py     # One-time super admin setup script
├── requirements.txt    # Python dependencies
├── .env.example        # Environment variables template
├── .env                # Your actual credentials (not in git)
└── uploads/            # Uploaded images (not in git)

MPF-frontend/
├── src/
│   ├── pages/
│   │   ├── PublicUpload.jsx        # Public sighting portal
│   │   ├── AdminLogin.jsx          # Organization login
│   │   ├── AdminDashboard.jsx      # Police/NGO dashboard
│   │   ├── Register.jsx            # Organization registration
│   │   ├── SuperAdminLogin.jsx     # Super admin login
│   │   └── SuperAdminDashboard.jsx # Approve/reject orgs
│   ├── components/
│   │   └── Navbar.jsx              # Navigation bar
│   └── App.js                      # Routes

---

🧠 How the AI Matching Works

1. Admin uploads missing person photo
         ↓
2. DeepFace extracts 512-dimension face encoding
   (VGG-Face model — trained on millions of faces)
         ↓
3. Encoding stored as JSON in MySQL
         ↓
4. Public uploads sighting photo
         ↓
5. System extracts encoding from sighting
         ↓
6. Cosine similarity calculated against ALL stored encodings
   similarity = dot(A, B) / (||A|| × ||B||)
         ↓
7. If distance < 0.4 threshold → MATCH FOUND
         ↓
8. Email sent to family with:
   - Match confidence percentage
   - Google Maps link of sighting location
   - Actual sighting photo attached

---

📧 Email Alert Preview

When a match is found, family receives:

🚨 POSSIBLE MATCH FOUND

Person          : [Name]
Match Confidence: 87.5%
Location        : 12.9716, 77.5946

[ 📍 View Location on Google Maps ]

📸 Photo uploaded by public: [image attached]

⚠️ This is an AI-based match. Please verify in person.

---

🔮 Future Improvements

• Deploy on AWS EC2 + S3 for image storage
• Super Admin web portal (currently API only)
• WhatsApp alerts via Twilio API
• Video frame extraction for sighting videos
• Mobile app (React Native)
• National missing persons database integration

---

👨‍💻 Author

---

Built with ❤️ to help reunite families