<![CDATA[Indonesia is one of the countries facing serious problems related to the high number of smokers. Active smokers have a high risk of contracting various serious diseases, such as heart disease, cancer, respiratory diseases, and others. Additionally, exposure to tobacco smoke also has adverse effects on passive smokers, who are often individuals around them who do not smoke but are affected by it. Conventional methods for detecting smokers are often inefficient and require significant manual intervention, thus necessitating a technological solution for automatic and real-time detection to support the enforcement of anti-smoking regulations. Therefore, this research aims to detect smoking behavior using the You Only Look Once (YOLO) version 8 method on Apple M2. YOLO V8 was chosen for its capability in fast and accurate object detection, while the Apple M2 supports real-time processing. The training results showed an accuracy rate of 91.6%, precision of 96.4%, recall of 90.4%, and an F1-Score of 93.2%. During the inference stage, the Apple Neural Engine (ANE) was able to process 21-25 frames per second (fps), demonstrating good capability for real-time object detection. The combination of YOLO V8 and Apple M2 proved effective for detecting smokers in public areas, offering an efficient and effective innovative solution, supporting the creation of a smoke-free environment in Indonesia, and showing great potential for the application of edge computing in similar applications in the future.]]>
Copyrights © 2024
