Social Distancing Analyzer

Abstract

With the spread of Corona Virus, a disastrous and infectious disease due to which there was loss of millions of lives and is still continuing to be a threat to human life. One of the effective measures that helped in stopping the transmission to some extent is social distancing, which is often known as physical distancing. It has been acknowledged internationally as a non- drug avoidance measure for COVID-19 transmission. This work proposes a system named Social Distancing Analyzer that can be used for alerting people to maintain a safe distance to stop the spread of disease with each other.  The proposed system uses the video frame from camera calibrated into bird’s view as input which is then used for person detection using YOLOv3 algorithm, an open-source object detection model. A centroid can be computed by following the subjects and calculating the distance between individuals. The distance obtained by the system is then used to determine whether the rules are being followed   or not. If the distance between two people is above minimum threshold value, the individual is represented using green box which also mean that social distancing is being followed, if not then it is represented using red box and an alert is generated showing that social distancing is not being followed.