A complete end-to-end AI-driven solution developed at the FloodAI Hackathon 2024 by Team Epsilon. Our system focuses on flood prediction, risk assessment, satellite image analysis, and real-time alert generation for Mumbai using meteorological, topographical, and urban data sources.
-
📊 Rainfall Prediction Model:
Time series model trained on IMD data to forecast rainfall in mm. -
🗺️ Flooded Area Segmentation:
Semantic segmentation using satellite images to identify flooded regions. -
🌧️ Risk Assessment System:
Combines rainfall, elevation, drainage network, and land use data to compute flood risk across Mumbai. -
📍 Interactive Map & Alert System:
Web-based dashboard to visualize risk zones and push alerts to residents. -
🧠 Integrated AI Pipeline:
Combines predictive models, geospatial analysis, and real-time user interaction.
🥉 Secured 3rd place at FloodAI Hackathon organized by IIT Gandhinagar and IIT Bombay Climate Studies Department.
🔬 Developed a research-backed multi-model architecture within 36 hours.
📍 Mumbai-focused system with real-world applicability.
- Languages: Python, JavaScript (for dashboard)
- Libraries: TensorFlow, Keras, OpenCV, Scikit-learn, XGBoost, Pandas, NumPy, Matplotlib, Rasterio
- Geospatial Tools: QGIS, Folium, GeoPandas, Shapely
- Frontend: Leaflet.js, Streamlit
- Other: Google Earth Engine, SRTM DEM, IMD Rainfall Data
- Clone the repository
git clone https://github.com/DeepMathukiya/FloodAI_Hackathon_Epsilon.git
cd FloodAI_Hackathon_Epsilon- Run Backend
cd backend
python app.py- Run Frontend
cd frontend
yarn run dev --host