Implementation of YOLO and Smoke Sensor for Automating Public Service Announcement of Cigarette's Hazard in Public Facilities

Smoking in public facilities is one of the most unwanted behaviours that shall be given serious and continuous approaches. Using automation-based approaches, nowadays, a common technique used is using detectors which trigger an alarm upon detecting smoke. While this approach is better than none, it is more detective than preventive and is still lacking awareness-stimulation. This research aims to propose a more comprehensive solution. A real-time alert is not only triggered when the system detects a certain level of smoke, but also when it recognizes cigarette objects, to add the preventive purpose. Not only a buzzer is triggered, upon detection of either smoke or cigarette or both, a public service announcement (PSA) is also automatically displayed to provide education about smoking's hazard to conduct responsible behavior in public facilities. The system combines the concepts of embedded systems and deep learning. The process consists of a continuous monitoring of smoke level from several sensors reading and also a continuous cigarette object detection using YOLO. The outputs are two binary digits that will role as the inputs for an OR operation which results in the bit required to trigger an anti-smoking PSA and a buzzer. The system consists of a Raspberry Pi 4B 4GB, smoke sensors, a camera module as the visual input device, a monitor, and also a buzzer. The results show that the system is able to perform in a semi indoor environment for detecting both smoke and cigarette then automating an alert and a PSA in a soft real-time manner.