Gaussian Process for the Machine Learning-based Smart fire Detection System

Abstract

Smart fire detection systems should be able to detect the fire and trigger the automatic alarm at an early stage. It should also trigger the automatic fire extinguishing system and broadcast the fire alarm under different fire conditions. Due to the strict detection accuracy requirement of the fire detection system, most of the modern smart fire detection systems are based on multi-sensor, or image/video surveillance system to reinforce its fast reaction and high reliability in the action process. In this paper, the multi-sensor detection system is combined with image recognition process. Image recognition is utilized to help the fire detection, when the decision from the multi-sensor system is uncertain or the data is not available/faulty. Image features are extracted by using machine learning methods. Then, the Gaussian classification method is applied to detect the specific fire case. Images from real environments are used to evaluate the proposed method. In addition, we investigate and discuss the detection results when the training data is adequate or inadequate, which verifies that the image-based fire detection scheme combined with multi-sensor system can achieve better accuracy.