Note: This project was built quickly during a national level hackathon (achieving 3rd rank). The original repository crashed, so it was moved here for presentation purposes. Commit history may be sparse.

NETRA — Autonomous Pothole Intelligence Platform

Neural Engine for Terrain Recognition & Analysis

---

🚀 What is NETRA?

NETRA is a real-time, edge-optimized AI platform for detecting, segmenting, measuring, and triaging road potholes and anomalies from dash-cam video, drone footage, and citizen uploads. It combines deep learning, predictive analytics, and interactive web dashboards to transform road safety and maintenance.

---

🧠 Core Features

• Deep Learning Dashcam Pipeline
  • YOLOv8 instance segmentation for pothole/crack detection
  • MiDaS monocular depth estimation for severity scoring
  • Temporal tracking and repair verification
  • 3D mesh and point cloud export
• Predictive Early-Warning Engine
  • Live weather vulnerability index (OpenWeatherMap API)
  • Degradation risk forecasting and ROI metrics
• Interactive Web Dashboard
  • React + Vite frontend with live maps, analytics, and citizen portal
  • Persistent AI Copilot (Google Gemini LLM integration)
• API & Data Sync
  • Node.js REST API for pothole records, complaint tracking, and CPGRAMS sync
  • MongoDB backend
• Synthetic Demo Runner
  • Generate synthetic road frames for end-to-end pipeline testing

---

🏗️ Architecture Overview

+-------------------+      +-------------------+      +-------------------+
|   Dashcam Video   | ---> |   AI Pipeline     | ---> |   Web Dashboard   |
+-------------------+      +-------------------+      +-------------------+
        |                        |                        |
        v                        v                        v
  YOLOv8 Segmentation   MiDaS Depth Estimation   React + Gemini Copilot
        |                        |                        |
        v                        v                        v
  Temporal Tracking     Severity Scoring         Live Map & Analytics
        |                        |                        |
        v                        v                        v
  GPS Geotagging        Repair Verification      Citizen Portal
        |                        |                        |
        v                        v                        v
  3D Mesh Export        API Sync & Alerts        Complaint Tracker

---

🏁 Quick Start

1. Install Python dependencies:

   pip install -r NETRA-AI/requirements.txt

2. Run the AI pipeline on a video:

   cd NETRA-AI
   python pipeline.py --source ../video/sample.mp4

3. Launch the web dashboard:

   cd netra-web
   npm install
   npm run dev

4. Start the API server:

   cd netra-api
   npm install
   npm start

---

🧩 Project Structure

• NETRA-AI/ — Python AI pipeline, models, demo scripts
• netra-web/ — React + Vite frontend, live map, analytics, chatbot
• netra-api/ — Node.js REST API, MongoDB integration
• archive/ — CSV datasets for accidents, vehicles, casualties
• video/ — Sample videos, deep-dive docs

---

🌟 Unique Selling Points

• Multi-format ingestion: video, images, citizen uploads
• Automated PDF report generation for contractors
• Proactive asset management with predictive alerts
• Conversational AI Copilot for admin and citizen support
• 3D mesh export for drone micro-scan workflows

---

📚 Documentation & Demos

• Deep Dive Architecture & Features
• AI Pipeline Details
• Web Dashboard Features

---

🤝 Contributing

We welcome contributions! Please open issues, pull requests, or join our discussion forum for feature requests and bug reports.

---

🏆 Credits

• Developed by the NETRA Team
• Powered by YOLOv8, MiDaS, React, Node.js, MongoDB, Gemini LLM

---

Ready to make roads safer, smarter, and more resilient? Dive in and run NETRA today!

---