MACHINE LEARNING-BASED SURVEILLANCE FOR FIRE ACCIDENT MONITORING

Abstract

In the face of diverse environments ranging from sprawling urban landscapes to dense jungles,
the threat posed by fire accidents remains a significant concern worldwide. Despite the potential for mitigating
such risks through the deployment of fire detection systems, existing solutions are hindered by several
challenges. These include prohibitive costs, high rates of false alarms, the necessity for dedicated infrastructure,
and the overall lack of robustness in both hardware and software-based detection systems. To address these
obstacles, our research endeavors to advance fire detection capabilities harnessing the power of deep learning
techniques, particularly within the domain of video analysis. Deep learning, rooted in artificial neural networks,
has demonstrated remarkable across various disciplines, notably in computer vision tasks. Our proposed system
is designed to detect fires in videos swiftly and reliably, with the capability to operate effectively across diverse
environments and sends messages to the owner