RoadPulse AI is an AI-powered platform that enables citizens to report road infrastructure issues such as potholes, waterlogging, accidents, blocked roads, and traffic signal failures. The system automatically analyzes uploaded images, classifies incidents, detects duplicate reports, assigns the appropriate government department, and stores reports for monitoring and analytics.
This project was built as an MVP for an Agentic AI Hackathon.
Road infrastructure issues often remain unreported or are reported through informal channels.
As a result:
- Delayed repairs
- Vehicle damage
- Traffic congestion
- Increased accident risk
- Lack of transparency
RoadPulse AI solves this by creating a community-driven reporting platform powered by AI.
- Upload road issue images
- AI-powered image analysis
- Automatic incident classification
- Severity prediction
- Duplicate report detection using embeddings
- Automatic department routing
- Report management dashboard
- Analytics dashboard
- SQLite database
- REST API using FastAPI
- Modern Streamlit frontend
- Pothole
- Waterlogging
- Accident
- Blocked Road
- Signal Failure
- Construction
- Normal Road
Citizen
│
▼
Upload Image + Location
│
▼
Streamlit Frontend
│
▼
FastAPI Backend
│
▼
AI Processing Pipeline
┌──────────────┬──────────────┬─────────────┐
│ │ │ │
▼ ▼ ▼ ▼
Image AI Embedding Duplicate Routing
Analysis Generation Detection Department
└──────────────┬──────────────┬─────────────┘
▼
SQLite Database
│
▼
Reports + Dashboard + Analytics
- Streamlit
- Plotly
- Folium
- Requests
- FastAPI
- SQLAlchemy
- SQLite
- Pydantic
- Google Gemini Vision (Image Understanding)
- Groq LLM
- Sentence Transformers
- Cosine Similarity
- Render
- GitHub
RoadPulse-AI/
│
├── frontend/
│ └── app.py
│
├── api/
│ └── routes.py
│
├── database/
│ ├── db.py
│ ├── crud.py
│ └── models.py
│
├── schemas/
│ └── report.py
│
├── services/
│ ├── gemini.py
│ ├── embedding.py
│ ├── duplicate.py
│ ├── routing.py
│ └── pipeline.py
│
├── static/
│ └── uploads/
│
├── main.py
│
├── requirements.txt
│
└── README.md
POST /reports
│
▼
Upload Image
│
▼
Analyze Image
│
▼
Generate Embedding
│
▼
Check Duplicate
│
▼
Assign Department
│
▼
Save Database
│
▼
Return JSON Response
The dashboard provides:
- Total reports
- Pending complaints
- Resolved complaints
- Critical incidents
- Recent activity
- AI insights
- Incident Distribution
- Severity Analysis
- Department-wise Complaints
- Resolution Status
- Infrastructure Statistics
Clone the repository
git clone https://github.com/yourusername/RoadPulse-AI.gitMove into the project
cd RoadPulse-AICreate a virtual environment
python -m venv venvActivate virtual environment
venv\Scripts\activatesource venv/bin/activateInstall dependencies
pip install -r requirements.txtCreate a .env file
GROQ_API_KEY=YOUR_GROQ_API_KEYuvicorn main:app --reloadOpen
http://127.0.0.1:8000/docs
streamlit run frontend/app.pyIf more development time is available, RoadPulse AI can be extended with:
- Real-time traffic prediction
- Live alternate route recommendation
- GPS-based automatic location detection
- Government complaint portal integration
- WhatsApp and Telegram reporting
- Push notifications
- Ward-level infrastructure quality score
- AI-based complaint verification from multiple users
- Multi-language support
- Real-time admin dashboard
- Cloud deployment with PostgreSQL
- Authentication and user profiles
RoadPulse AI combines Artificial Intelligence, Computer Vision, GIS, and Community Reporting to create a smarter, safer, and more responsive road infrastructure management system.
Instead of simply collecting complaints, the platform analyzes, validates, prioritizes, and routes issues automatically, reducing manual effort and enabling faster action.
Developed for the Agentic AI Hackathon. Developed by Gargee Sharma
This project is intended for educational and hackathon purposes.