Flood Segmentation Platform
A real-time flood monitoring platform using Sentinel-1 SAR satellite imagery and a U-Net deep learning model for flood segmentation.
Problem
Flood monitoring in Nepal relies on manual reporting and delayed satellite data.
Solution
An end-to-end platform that ingests Sentinel-1 SAR data, runs inference through a U-Net model, and delivers flood segmentation maps with real-time alerts via WebSockets.
Key Features
U-Net deep learning model for satellite image flood segmentation
Containerized architecture with React, Node.js, and FastAPI services
WebSocket-based disaster alert delivery
Geospatial flood visualization using Leaflet
Outcomes
Processes satellite imagery for automated flood mappingReal-time alert delivery to monitored regions
Tech Stack
ReactTypeScriptNode.jsExpressMongoDBSocket.IOFastAPIPyTorchLeafletDocker