Forest Fire Detection: A Comparative Analysis of Deep Learning Algorithms
M. Karthi, R. Priscilla, G Subhashini, Niroshini Infantia C, G R Abijith, J Vinisha
Abstract
Fires are a serious hazard to the planet, from spreading cities to impenetrable jungles. These must be avoided by implementing fire detection systems, however, the high cost, specialized connectivity, misfires, and lack of reliability of current facilities-based detection systems have functioned as hurdles. In this paper, we use Deep Learning to make a step toward detecting fire in videos. Deep learning is a novel concept built on artificial neural networks where it has excelled in a number of fields. We intend to resolve the inadequacies of current methods and develop a precise and perceptive system that can identify firesas quickly as practicableand function in a range of environments, saving a tremendous amount of time and resources. the recommended commended a fire detection is an approach based on Convolutional Neural Networks-CNN and the YOLOV3 method to accomplish well ad suitablefire picture detection, which is compatible with fire detection through dataset training. existing fire-detection systems in smart cities can be changed with computer vision techniques to build fire safety in the community using digital innovations. In this work, we created a fire detector that can quickly and precisely recognize even the smallest sparks and sound an alarm if a fire starts over eight seconds. An improved You Only Look Once (YOLOv3) network was used to create a revolutionary convolutional neural network that can detect fire zones.