Namma AI

Smarter Roads. Safer Journeys.

Traffic Intelligence Platform for Bengaluru

---

Overview

Namma AI helps Bengaluru Traffic Police shift from reactive patrol to proactive, data-driven enforcement. It predicts incident severity and road-closure probability using a CatBoost ML model, recommends officer and barricade deployment, generates diversion routes, surfaces hotspots on a live heatmap, and lets citizens report incidents directly to authorities.

Next.js 14 · FastAPI · PostgreSQL · CatBoost

Key Metrics

Metric	Value
ML inference latency	< 5 ms per prediction
Diversion route computation	< 3 s
Incident telemetry	Real-time via WebSocket
Authentication	JWT-based
Access control	Role-based (Citizen / Personnel / Authority)
API	REST, with auto-generated Swagger documentation

---

Tech Stack

Layer	Technology
Frontend	Next.js 14, TypeScript, Tailwind CSS, Leaflet (OpenStreetMap), Recharts
Backend	FastAPI, Pydantic v2, SQLAlchemy
Database	PostgreSQL
ML	CatBoost (closure probability + priority ranking), DBSCAN (hotspot clustering)
Routing	OSRM public server + custom diversion engine
Maps & weather	OpenStreetMap tiles, Open-Meteo API (both free, no key needed)

---

---

Folder Structure

namma-traffic/
├── backend/          # FastAPI app, models, routers, ML inference
│   ├── models/       # CatBoost + DBSCAN model artifacts
│   ├── routers/      # API route handlers
│   └── main.py
├── frontend/         # Next.js 14 app (citizen + authority views)
│   ├── app/
│   └── components/
├── docs/             # Screenshots, logo, hero images
└── README.md

---

Run Locally

Prerequisites

• Python 3.12+ — tick "Add python.exe to PATH"
• Node.js LTS
• PostgreSQL 14+

1 — Database

createdb traffic_db

2 — Backend

Open a terminal in the project root:

# Windows (PowerShell)
cd backend
py -3.12 -m venv venv
venv\Scripts\Activate.ps1
pip install -r requirements.txt
uvicorn main:app --reload

# Mac / Linux
cd backend
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
uvicorn main:app --reload

If PowerShell blocks script execution, run once: Set-ExecutionPolicy -Scope CurrentUser -ExecutionPolicy RemoteSigned

Backend: http://localhost:8000 · Swagger docs: http://localhost:8000/docs

The schema (all tables and columns) is created automatically on first start.

3 — Environment variables

Create backend/.env:

DATABASE_URL=postgresql://user:pass@localhost:5432/traffic_db
JWT_SECRET_KEY=any-random-string-here

# Optional — only needed for Gemini advisory and Cloudinary photo upload
GEMINI_API_KEY=
CLOUDINARY_CLOUD_NAME=
CLOUDINARY_API_KEY=
CLOUDINARY_API_SECRET=

4 — Frontend

Open a second terminal:

cd frontend          # or: cd namma-traffic/frontend
npm install
npm run dev

Open http://localhost:3000

---

Demo Flow

1. Open http://localhost:3000
2. Seed data — click "I'm an Authority" → log in (admin / admin123) → Dashboard → expand Admin Tools → click Generate demo data
3. Citizen view — go back to home → click "I'm a Citizen" → try Safe Route Finder and Live Heatmap
4. Report an incident — Citizen → Report Incident → fill form → submit → copy tracking ID
5. Verify it — Authority → Verify Reports → approve the report → watch it appear on the dashboard live
6. Resource planning — Authority → Deployment → see per-incident officer and barricade recommendations
7. Diversion — Authority → Diversion Plans → click "Diversion plan" on any incident

---

Roles & Logins

Role	How to access	Credentials
Citizen	Home → "I'm a Citizen"	No login needed
Traffic Personnel	Home → "I'm Traffic Personnel"	Register or use demo login
Authority	Home → "I'm an Authority" → Login	admin / admin123

---

Page Routes

Route	Who	Description
/	Public	Landing + role selection
/citizen/report	Citizen	Submit incident report
/citizen/heatmap	Citizen	Live incident map + hotspots
/citizen/route	Citizen	Incident-aware safe route planner
/authority/login	Authority	JWT login
/authority/dashboard	Authority	Live KPIs + priority incidents
/authority/predict	Authority	ML closure probability + priority ranking
/authority/analytics	Authority	Incident trends (charts from live DB)
/authority/resources	Authority	Per-incident deployment recommendations
/authority/diversion	Authority	Auto-generated diversion routes
/authority/verify	Authority	Approve / reject citizen reports

---

API Example

POST /api/predict — returns road-closure probability and priority ranking for an incident.

Request:

{
  "incident_type": "accident",
  "location": {
    "lat": 12.9716,
    "lng": 77.5946
  },
  "severity": "moderate",
  "weather": "rain",
  "time_of_day": "peak_hour",
  "road_type": "arterial"
}

Response:

{
  "closure_probability": 0.72,
  "priority_score": 8.4,
  "recommended_officers": 3,
  "recommended_barricades": 2,
  "estimated_clearance_minutes": 35
}

Full interactive docs for all endpoints are available at /docs (Swagger UI) once the backend is running.

Deploy

• Backend → Render: New Web Service · root backend · start command uvicorn main:app --host 0.0.0.0 --port $PORT · set env vars in Render dashboard
• Frontend → Vercel: Import repo · root frontend · set NEXT_PUBLIC_API_URL to your Render backend URL

---

Namma Traffic · Smarter Roads. Safer Journeys.
BIT Mesra