Robust Real-Time Fire Detector Using CNN And LSTM

Abstract

The Detection of a fire in surveillance systems is playing a significant role to Reduce material and human losses, the effectiveness of fire detectors measured by the speed of response and the accuracy and the generality over different kinds of video sources with a different format. Several studies worked on fire detection. Also, there is several benchmarking dataset, even though all available datasets are not large enough to build a robust real-world fire detector. In this paper, we proposed a real-time fire detector based on deep-learning, the model consists of Convolutional neural network (CNN) as spatial feature extractor and Long short-term memory (LSTM) as temporal relation learning method with a focus on the three-factor (overall generality - accuracy - fast response time) the proposed model achieved accuracy of 95.39% with a speed of 120 frames/sec based on extended dataset from the available data sets. The accuracy and the speed of the proposed model have been compared to previous works, shows that the proposed model has the highest accuracy and the fastest speed among all the previous works in the field of fire detection.