DamageSense AI

DamageSense AI is a Flask dashboard for rapid structural damage assessment. Users can register, log in, upload or capture building images, and keep assessment history private to their own account.

What It Does

• User registration and login
• Private per-user assessment dashboards
• Image upload assessment
• Live camera capture assessment on HTTPS deployments
• Server-backed assessment history
• Optional professional hosted model integration through Roboflow
• Local lightweight image-analysis fallback when no hosted model is configured

Tech Stack

Layer	Technology
Backend	Python, Flask 3
Database	SQLAlchemy, SQLite
Image Processing	Pillow, NumPy
Optional Hosted AI	Roboflow hosted detection/segmentation model
Frontend	HTML, Bootstrap, browser camera APIs
Testing	Pytest
Deployment	Gunicorn, Render.com

Project Structure

DamageSense-AI/
├── app/
│   ├── __init__.py
│   ├── routes.py
│   ├── models.py
│   ├── ai_engine.py
│   └── static/uploads/
├── templates/
│   ├── login.html
│   ├── register.html
│   └── upload.html
├── tests/test_app.py
├── .env.example
├── requirements.txt
├── run.py
└── README.md

Local Setup

git clone https://github.com/WIZZ249/DamageSense-AI.git
cd DamageSense-AI
python -m venv venv
venv\Scripts\activate
pip install -r requirements.txt
cp .env.example .env
python run.py

Open http://localhost:5000.

API Endpoints

Method	Endpoint	Description
GET	/	Redirects to login or user home
GET/POST	/register	Create a user account
GET/POST	/login	Log in
GET	/logout	Log out
GET	/home	User dashboard
POST	/assess	Upload or camera-captured image assessment
GET	/history	Current user's latest 50 assessments
GET	/health	Health check

Professional Model Setup

The app works immediately with a local Pillow/NumPy fallback, but professional damage assessment should use a trained damage model.

Recommended path:

1. Train or choose a Roboflow hosted model for structural/building damage detection.
2. Use damage classes such as no_damage, minor_damage, major_damage, and destroyed.
3. In Render, add these environment variables:

ROBOFLOW_API_KEY=your_api_key
ROBOFLOW_MODEL_ID=your-project/version
ROBOFLOW_CONFIDENCE=35
ROBOFLOW_OVERLAP=30

When these are present, app/ai_engine.py calls the hosted model first. If the hosted call fails or is not configured, the app falls back to local image analysis so uploads still work.

Recommended datasets and model families:

• RescueNet for high-resolution UAV disaster imagery
• xBD/xView2 for satellite before/after disaster damage assessment
• YOLO segmentation for fast object-level field use
• SegFormer for stronger semantic segmentation quality

Camera Capture

The dashboard includes a Take Picture mode. Browser camera access requires HTTPS, which Render provides on deployed services. On phones, the upload field also hints to open the device camera.

Running Tests

pytest tests/

Deployment on Render

1. Connect this repository to Render.
2. Use start command:

gunicorn run:app --timeout 120 --workers 1

3. Add environment variables from .env.example.
4. Deploy from main.

For serious production use, replace SQLite with PostgreSQL so user accounts and assessment history persist reliably across restarts.

License

MIT