Research on Cable Fire Detection and Early Warning Based on Deep Learning

Abstract

Fire detection based on computer vision is an important task in modern monitoring systems. In recent years, due to the high-precision recognition rate of Convolutional Neural Networks (CNN), applications in the field of fire detection and recognition based on Convolutional Neural Networks (CNN) have continued to develop. In this paper, aiming at the realistic requirements of cable fire, we build an algorithm based on flame detection, process the picture data of multiple scenes monitored by ordinary cameras, and use the deep neural network model based on CNN to classify, identify and detect the pictures with fire/smoke and without fire. Finally, according to the classification results, when there is a fire alarm signal is displayed. Experimental results show that this method can effectively identify cable fires, achieve a higher recognition and alarm rate in the self-made database, meet the expected functions and index requirements, and achieve the goal of cable fire warning.