ABSTRACT
The progressing COVID-19 flare-up has caused a worldwide calamity with its lethal spreading. The infection spreads rapidly and is a threat to humankind. Seeing the need of great importance, one should consistently play it safe of which one being social distancing. Social distancing is considered to be the most suitable measure against the powerful COVID-19 affliction. As per the WHO, to practise appropriate social distancing, individuals should keep 3ft or 1m distance between one another. This framework centres around an answer for determining social distancing utilizing YOLO object discovery on video film and pictures continuously. The system utilizes the YOLOv5 object detection to distinguish people in a video. The detection model distinguishes people groups utilizing recognized bounding box data. The pairwise distances of the centroid of the distinguished bounding boxes of the individuals are resolved using the Euclidean distance. To detect the social distance infringement between individuals, we are utilizing an estimate of the actual distance to pixel and setting and edge. An infringement edge is set up to assess whether the distance esteem breaks the base social distance limit. The framework ensures higher derivation speed and is subsequently fit for conveying real-time outcomes withoutlosing on precision, even in much complex arrangements. © 2021 Turkish Physiotherapy Association. All rights reserved.