MobileNet SSDv2 Inference Approach of Smoke Hazard Detection and Alert System: A Smoke-Induced Simulated Home-Environment

Abstract

A smoke detector is an instrument that detects smoke usually as a fire warning. The devices available are placed in the ceiling and they take too long to respond to smoke. These smoke alarms are helpful when the smoke is big enough to reach the ceiling and alarm that there is a fire in the area. It takes a big fire and a lot of smoke before it can be detected and before the alarm goes on. In this study, the detection of smoke will be done by using an object-detection algorithm. It detects smoke early even if there is only a small amount to be detected. These detected smokes are then analyzed and will then inform the user about the detected smoke. This study can help a lot in fire prevention because of the detection of smoke inside the house and can prevent fire as early as the smoke has been recognized by the camera. The system produced an overall 89% testing accuracy.