Unlocking Doors: A TinyML-based Approach for real-time Face Mask Detection in Door Lock Systems

Abstract

In response to the rapid spread of coronaviruses, including COVID-19 and seasonal common cold viruses, this article introduces a proposed system for enhancing door lock systems using TinyML technology for real-time face mask detection. The research project focuses on developing a machine learning model based on the YOLOv5 architecture to classify individuals based on their mask-wearing behavior correctly, incorrectly, or not at all in high-risk spaces prone to the transmission of coronaviruses, such as healthcare facilities, laboratories, and public settings. The study outlines the hardware and software tools utilized, including the Raspberry Pi 4, camera hardware, and the YOLOv5 machine learning model. The model is trained using a dataset containing three different classes and converted to a TFLite format for efficient implementation on the Raspberry Pi. Evaluation results demonstrate a mean Average Precision (mAP) of 0.99 and an inference rate of 10FPS for a 128-frame size input. This proposed system offers practical implications for enhancing door lock systems and promoting public health and safety during outbreaks of coronaviruses, including COVID-19 and other seasonal coronaviruses, providing a valuable approach to decrease the spread of these diseases and mitigate transmission risks in high-risk spaces, thereby contributing to the overall reduction of public health threats.