Sistem Deteksi Orang Merokok Menggunakan YOLO11 dengan Pemberitahuan Otomatis Via Telegram

Abstract

Indonesia ranks fourth globally in terms of the highest number of smokers as of 2025. Exposure to cigarette smoke poses health risks not only to active smokers but also adversely affects passive smokers due to the toxic substances contained in the smoke. To address this issue, the government has enacted Law Number 36 of 2009 concerning Smoke-Free Areas (Kawasan Tanpa Rokok). However, its implementation remains hindered by ineffective and limited manual monitoring. This research aims to develop a system for detecting individuals who are smoking by utilizing the You Only Look Once version 11 (YOLO11). The system is designed to automatically detect cigarettes, followed by triggering a voice warning and sending instant notifications through a Telegram Bot. The model was trained using a dataset of cigarette images and evaluated to measure its effectiveness in object detection. The evaluation results show that the system achieved an accuracy of 85%, with a precision of 90%, recall of 92%, and F1-score of 90%. In real-time testing, the system demonstrated an average response time of 39.30 milliseconds and a processing speed of 7.20 frames per second, indicating relatively responsive performance. However, certain factors such as visual obstructions, suboptimal image quality, and object similarities still affect detection accuracy.