Nayan: Protecting Women from Safety Threats

Nayan is an AI-powered safety ecosystem designed to enhance women's safety in urban environments. By integrating real-time threat detection, automated alerts, and a community-based safety network, it bridges the gap between citizens and emergency services. The system leverages advanced Computer Vision, Machine Learning, and IoT technologies to create a proactive safety net.

🚀 Key Features

• Real-time Threat Detection: AI models analyze live video feeds to detect SOS gestures (e.g., open palm, tucked thumb) and potential violence.
• Smart Surveillance: Automated gender classification and crowd density analysis (Gender Ratio) to identify potential risk zones in real-time.
• Hotspot Mapping: Dynamic identifying of high-risk areas based on historical data and live analytics, alerting users when they enter these zones.
• Emergency SOS: One-touch activation that triggers:
  • Instant SMS/Call to emergency contacts.
  • Live location sharing with law enforcement (112 integration).
  • Visual and audio alerts on the monitoring dashboard.
• Safe Mode: Continuous background monitoring that tracks location and status, automatically triggering alerts if anomalies are detected (e.g., prolonged inactivity, route deviation).
• Interactive Dashboard: A centralized command center for authorities to view live camera feeds, track active SOS alerts on a map, and manage resources.
• Mobile App: A user-friendly application for women to access safety tools, view nearby safe zones (hospitals, police stations), and report incidents.

🏗️ System Architecture

The system operates on a client-server model where multiple video sources (Mobile Phones, CCTV) feed data to a central processing unit. The analyzed data is then distributed to the Command Center Dashboard and the User Application.

graph TD
    subgraph "Data Sources"
        CCTV[CCTV Cameras] -->|RTSP Stream| Backend
        MobileApp[User Mobile App] -->|Video/Location/SOS| Backend
    end

    subgraph "Backend Core (FastAPI)"
        Backend[FastAPI Server]
        ML_Engine[AI/ML Engine]
        DB[(Database - Supabase/Postgres)]
        
        Backend <--> ML_Engine
        Backend <--> DB
    end

    subgraph "AI/ML Engine"
        YOLO[YOLOv8 Detection]
        MediaPipe[MediaPipe Gesture]
        OpenCV[OpenCV Processing]
        
        ML_Engine --> YOLO
        ML_Engine --> MediaPipe
        ML_Engine --> OpenCV
    end

    subgraph "Clients"
        Dashboard["Admin Dashboard (React)"]
        Police["Emergency Services (112)"]
        User["User App (React Native)"]
    end

    Backend -->|WebSocket/Alerts| Dashboard
    Backend -->|SMS/API| Police
    Backend -->|Notifications| User

Loading

🛠️ Technology Stack

Component	Technology	Description
Backend	Python, FastAPI	High-performance API server handling requests and WebSocket connections.
AI/ML	YOLOv8, MediaPipe, OpenCV	Real-time object detection, gesture recognition, and image processing.
Database	Supabase (PostgreSQL)	Stores user data, alert logs, and camera configurations.
Mobile App	React Native (Expo)	Cross-platform mobile application for Android and iOS.
Frontend	React.js, Bootstrap	Admin dashboard for monitoring and analytics.
Maps	Leaflet, Google Maps API	Visualization of hotspots, cameras, and live tracking.
Communication	WebSocket, Twilio	Real-time data streaming and SMS/Call dispatch.

🔄 Data Flow

This sequence reflects the lifecycle of an SOS event or Threat Detection.

sequenceDiagram
    participant P as Public/User (Camera)
    participant B as Backend Server
    participant AI as MI Processor
    participant D as Dashboard (Admin)
    participant E as Emergency Contacts

    P->>B: Stream Video / Trigger SOS
    B->>AI: Send Frame for Analysis
    AI-->>B: Return Detection Result (Violence/Gesture)
    
    rect rgb(255, 230, 230)
        Note over B, D: If Threat Detected
        B->>B: Log Alert in Database
        B->>D: Push Real-time Alert (WebSocket)
        B->>E: Send SMS/Call (Twilio)
    end
    
    D->>D: Flash Warning & Show Location

Loading

📱 Use Cases

graph LR
    User((User/Public))
    Admin((Admin/Police))
    
    subgraph "Nayan System"
        SOS[Trigger SOS]
        SafeRoute[View Safe Route]
        ViewHotspots[View Hazard Hotspots]
        Monitor[Monitor Live Feeds]
        ReceiveAlert[Receive Emergency Alert]
        ManageResource[Manage Cameras/Resources]
    end

    User --> SOS
    User --> SafeRoute
    User --> ViewHotspots
    
    Admin --> Monitor
    Admin --> ReceiveAlert
    Admin --> ManageResource
    
    SOS -.-> ReceiveAlert

Loading

⚙️ Installation & Setup

Prerequisites

• Node.js (v18+)
• Python (v3.9+)
• Expo CLI

1. Backend Setup

cd backend
# Create virtual environment (optional but recommended)
python -m venv venv
# Windows
.\venv\Scripts\activate
# Linux/Mac
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# Run the server
uvicorn main:app --reload

2. Dashboard Setup

cd dash_frontend
npm install
npm start
# accessible at http://localhost:3000

3. Mobile App Setup

cd app
npm install
npx expo start
# Scan the QR code with Expo Go app on your phone

👥 Team

• Aditya Sahu - AI/ML Lead
• Joy Banerjee - Cloud & DevOps
• Aman Hussain - Data Analyst
• Nausheen - Mobile App Developer

💡 Inspiration & Credits

We have drawn significant inspiration from Team Boka and their project Sakhi Sahayak. We are actively contributing to their original vision by transforming it into a fully production-ready product.

Our Key Contributions:

• Live Feed Integration: Moving beyond mock data to support real-time video processing from live cameras.
• Robust Backend Architecture: Re-engineering the backend to handle high-concurrency WebSocket streams and API requests ensuring reliability.
• Mobile-Web Integration: Seamlessly connecting the mobile user application with the centralized web dashboard for unified situational awareness.
• Latency Reduction: Contributing on reducing the latency in data transmission and processing.