YOLOv4-Based Monitoring Model for COVID-19 Social Distancing Control

ABSTRACT

The coronavirus disease 2019 (COVID-19) has appeared in December 2019 at Wuhan city, China. The virus started spreading over the world. Most of the governments have taken different measures to prevent the outbreak. Social distancing (SD) is one of the effective solutions to prevent the spread of COVID-19, in which people should maintain a specific distance between each other. This paper aims to provide a YOLOv4-based model for monitoring social distancing. The model begins by taking a video/picture as input and generating warnings of SD violation. The YOLOv4 we used in this model detects pedestrian’s people in public places such as streets, malls, train stations, and universities based on deep learning techniques. The model uses a predefined SD threshold (SDTH) and a violation index (VI) to determine when the violation occurs and trigger a warning sub-system to make an awareness action immediately. A comprehensive investigation and discussion on the existing literature of SD, object detection methods, and SD monitoring have also been provided in this paper. The model provided is supposed to operate continuously in the targeted places to monitor people, thus reducing the impact of COVID-19 spread. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.