AI-based Wildfire Detection Using Satellite Images and Drones
Loading...
Authors
Xiong, Xingguo
Gupta, Navarun
El-Sayed, Ahmed
Issue Date
2026-04-17
Type
Other
Language
en_US
Keywords
Artificial intelligence , Satellite images , Wildfires
Alternative Title
Abstract
With the increasing frequency and severity of wildfires, significant threats are posed to human life, infrastructure, and ecosystems. Wildfires cause widespread air pollution, property loss, and greenhouse gas emissions. Early detection is important for minimizing these impacts by timely response and effective resource deployment. This project utilizes a hybrid wildfire detection system that integrates machine learning analysis of NASA satellite imagery with real-time, drone based environmental sensing. Deep learning models based on YOLO (You Only Look Once) algorithms are trained for fire and smoke detection based on NASA satellite imagery. Once potential fire spots are detected, a drone can be deployed to collect localized information for further verification. The trained AI model is also deployed on edge AI devices such as Raspberry Pi 5 and NVIDIA Jetson Orin for drone-based visual monitoring. An Arduino-based multi-sensor platform featuring temperature, gas, and flame sensors enhanced localized fire detection. A custom alert system sends real-time email and SMS notifications upon detection. The system improves emergency response capabilities while supporting NASA’s climate resilience goals and providing STEM research opportunities for students.
Description
UB Rise 2026
Department of Electrical and Computer Engineering