Monitoring Social Distancing based on Regression Object detector for reducing Covid-19
Ruchi Jayaswal, Manish Dixit · 2022 IEEE 11th International Conference on Communication Systems and Network Technologies (CSNT) · 2022
To deal with the worldwide coronavirus epidemic situation, the procedure of pulling down the Covid-19 cases will be tough to handle, if people do not take steps to thwart the virus from spreading. One of the most significant strategies in this pandemic is to keep a safe distance among the persons in public areas. This paper aims to detect persons with social distance monitoring as a preventive technique in minimizing physical connection between persons. The workflow of this paper is to detect the people in the area of interest using the YOLOv5 model. The model is trained on the Open Images V6 dataset and takes categories of people and human faces. Further, a social distancing algorithm is applied to check the distance between two persons and mark the red boundary box if any person violates the rules. The network provides inference speed capable of offering real-time results with maximum accuracy. The accuracy is achieved 99% in real-time scenarios using the proposed work. In various contexts, the suggested social distancing strategy produces promising outcomes.