SafeStreet is an AI-powered platform designed to identify, assess, and prioritize road damage reports using computer vision, location metadata, and automated workflows. It consists of a mobile app for users and workers, and a web platform for authorities to manage and monitor infrastructure repair efficiently.
Poor road conditions are a major safety and economic concern. SafeStreet empowers local communities and authorities with:
- ✅ Real-time damage detection
- 🧠 Vision Transformer CNN + DETR model for classification and severity assessment
- 📊 Historical analysis and dashboard for authorities
- 🔔 Automated alerts and repair task generation
┌────────────┐ Upload Image + Metadata ┌────────────┐
│ Mobile │ ──────────────────────────────────▶ │ Backend │
│ (App) │ │ (Node.js) │
└────────────┘ └────┬───────┘
│
┌────────────┐ Review & Analysis + Status ▼
│ Authorities│ ◀──────────────────────────────┬┐ ┌────────────┐
│ (Website) │ └──▶│ Database │
└────────────┘ └────────────┘- App Users (Citizens/Workers): Upload road damage photos via the app
- Backend Server: Processes data, classifies severity, stores to MongoDB, and triggers alerts
- Authorities: Web dashboard to visualize and manage incoming reports
| Role | Access Interface | Capabilities |
|---|---|---|
| User | Mobile App | Upload image, view status, history |
| Worker | Mobile App | View assigned tasks, upload completion images |
| Authority | Web Dashboard | Review reports, assign priorities, track analytics |
- Upload road damage with images and location
- View status of submitted reports
- Track repair history
- View assigned repair tasks
- Upload evidence of completed work
- AI-driven prioritization based on severity
- Historical analytics & trends
- Model 1: CNN for road classification
- Model 2: DETR (DEtection TRansformer) for bounding box regression and segmentation
- Inference Pipeline:
- Classify Image (road or not road)
- Assess severity from bounding boxes (Low, Medium, High)
- Assign repair priority score
- Built with React Native (Expo SDK 53)
- Firebase Authentication
- Tab navigation via
expo-router - Libraries:
lucide-react-native,@react-native-async-storage/async-storage
- Developed with React.js
- Backend: Node.js + Express.js
- Database: MongoDB
- REST API with JWT authentication
- Real-time damage tracking
- Task progress and worker performance
- Sorting reports based on priority and severity
- Notification system for urgent cases
| Layer | Technology |
|---|---|
| Frontend (App) | React Native (Expo) |
| Frontend (Web) | React.js |
| Backend | Node.js, Express.js |
| Database | MongoDB |
| ML Models | CNN, DETR (PyTorch) |
| Auth | Firebase (App), JWT (Web) |
- User/Worker captures image → uploads via app
- Metadata auto-attached (location, timestamp)
- Backend API handles:
- Image preprocessing
- ML model inference
- Severity scoring and database update
- Authorities view new reports → assign tasks
- Workers complete tasks and upload results
- User gets notified once resolved
safestreet/
├── App_Dev/ # React Native frontend
├── Web_Dev/ # React.js frontend for authorities
├── server/ # Node.js + Express backend
├── main/ # CNN + DETR models and preprocessing
├── README.md # This file
This project is licensed under the MIT License.
Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.
For queries or collaboration:
- ✉️ Email: ozairgdsc@gmail.com , avanagantiabhinavreddy@gmail.com , krrkailasa@gmail.com
- 🔗 LinkedIn: https://www.linkedin.com/in/md-ozair-0a0241342/ , https://www.linkedin.com/in/avanaganti-abhinava-reddy/ , https://www.linkedin.com/in/kailasa-raghunandan-rao-968701286/
- 💻 GitHub: https://github.com/ozair-kmit
Empowering smarter streets with the power of AI — one pothole at a time!