Sistema para Monitoramento de Distanciamento Social Utilizando Câmeras Inteligentes

Abstract

The purpose of the Social Distancing Monitoring System Using Smart Cameras proposed in this work is to allow real-time monitoring of a space. This application can be used in a variety of situations, such as to reduce the movement of people in collective spaces, as well as allowing greater control in a given environment, thus avoiding agglo merations. In addition, this system can be useful for containing the spread of diseases that can be transmitted through the air, as happened with the disease caused by the SARS-CoV-2 virus, popularly known as coronavirus (DAS et al., 2021). From this, a system is designed and implemented with a monocular camera connected to a raspberry where a code capable of detecting people in an image and calculating the distance between them is executed. Furthermore, a web interface is also made available that allows ma naging devices as well as monitoring the information that is sent in real time by them. The results obtained reveal that the system achieved a detection rate of approximately 49% considering an Intersection Over Union of 50% between the bounding boxes added in pre-processing and those predicted by the model, for an Intersection Over Union of 50% at 95% the detection rate reaches 99%. The algorithm responsible for measuring the distance had an average absolute percentage error of 18.45%. These results demonstrate that the application can be viably used to help both public security forces and commercial purposes.