Final Year Project

Project Title

FEWMS - Farmevo Wildfire Monitoring System

Description

This project revolutionizes wildfire management by replacing slow, manual analysis with an automated, multi-modal deep learning pipeline. To overcome traditional barriers like smoke and clouds, we developed a dual U-Net architecture that fuses multispectral and radar satellite data. This allows the system to accurately map burn severity without requiring manual threshold tuning. By incorporating environmental factors like wind, elevation, and slope, our parallel model also forecasts fire spread probability for the next 24 hours. Finally, we integrated these complex predictions into an interactive, real-time dashboard. This translates the model outputs into highly actionable metrics for end-users, providing them with critical data such as estimated carbon emissions and immediate risks to local infrastructure.