Face Masks Usage Monitoring for Public Health Security using Computer Vision on Hardware

Wearing face masks is one of the direct measures that can help tackling the spread of the new coronavirus. In this paper we presented an architecture for face mask wearing detection using pre-trained deep learning models for computer vision and implementation of them on embedded hardware platforms. Three object detection models were fine-tuned and optimized to run on 4 different hardware platforms. The fine tuning and optimization of the models resulted in significant reduction of the inference time, thus making the use of this technology in IoT based security systems for real-time automatic monitoring of face masks wearing realisable.

I. Introduction

COVID-19 pandemic has appeared in December 2019 starting from China and started spreading in all over world and almost in every country of this world [1]. It has caused death to more than 8.8 million people worldwide with higher mortality rates in elders and vulnerable population. Governments have taken several measures like asking citizens to stay at home, to avoid the transportation and travel and general lockdown measures in order to stop the spread of the new coronavirus disease [2]. The need for isolation and lockdown have already caused severe negative impact to economies, thus making the need tackling the spread of the coronavirus urgent and essential.