Early Wildfire Detection Using CNN-Based VGG16 Model

Abstract

In 2024, more than 8,024 wildfires resulted in significant mortality and infrastructure destruction, necessitating effective early detection systems. Conventional approaches encounter obstacles such as protracted response times and data constraints. This research improves wildfire identification with a CNN-based VGG16 model, trained on the D-FIRE dataset utilizing data augmentation techniques. The model attained an accuracy of 97.5% and a minimal false negative rate, guaranteeing dependable early identification. AI-driven wildfire surveillance facilitates prompt intervention, mitigating wildfire effects and enhancing disaster management initiatives efficiently.