Detection of Social Distance and Intimation System for Covid-19

ABSTRACT

COVID-19’s is a novel corona virus, fast spread has caused substantial damage and affected more than tens of millions of individuals around the world. People frequently wear masks to safeguard themselves and others against the transmission of coronavirus. The world health organization conveys the people to follow the social distancing to prevent the spread of Covid. Researchers have proposed several machine learning models to classify the disease, but none have identified the algorithm which gives more accuracy. Also, similar studies that have proposed various other techniques for prediction. In addition to that maintaining social distancing is also a major factor. In these regions, personally monitoring whether individuals are maintaining social distancing or not is quite impossible. This study aims to develop a highly accurate and real-time technique for the automatic detection of individuals who are not maintaining social distancing. Three state-of-the-art object identification models, namely YOLOv4, Tiny-YOLOv4 are used to detect the objects. Many results suggest that YOLO v4 has the greatest mAP value of 88.90%, followed by YOLO v4 and Tiny-YOLO v4 with mAP values of 82.24% and 74.80%, respectively. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.