Deep Learning Approaches for Real-Time Fire and Smoke Detection in Forests: A Performance Analysis

The purpose of this paper is to analyze the application of Artificial Intelligence AI in the area of fighting forest fires, with a focus on the early detection of fire and smoke. To address the needs for real-time efficiency and effectiveness, various computer vision-oriented methods are investigated. Using VGG-16 network as a pre-trained convolutional neural network CNN; image classification takes place. Faster-R-CNN and YOLOV8 for fire and smoke detection. Overall, the performance assessments are performed in virtual 3D environment and the real world for determining the best suitable method that can be used in the real-time detection. This work is aimed at producing an output that can help devise strategies for better forest fire fighting and efficiency.