I. Introduction
As COVID-19 ravages on, life cannot stay at a standstill. Owing to this, life is returning to a new normal which includes COVID protocols as the new health and safety paradigm. Apart from wearing masks and sanitizing one’s hands and belongings, maintaining social distance from other people is an important and sure-shot way to avoid getting infected. With the reopening of many public places post lockdown with reduced capacities, it is not possible to constantly manually monitor or enforce these protocols. Hence the proposed system deals with the automated identification of individuals who are not following the social distance guidelines to reduce human effort and save time. We have used a Convolutional Neural Network and image transformation techniques to detect people and estimate the distance between them. A neural network is a hardware and/or software system modelled after the way neurons in the human brain work. It’s a group of algorithms that try to find relationships in a data set using a method that resembles the human brain’s operation. A convolutional neural network (CNN) is a form of artificial neural network that is specifically built to recognise and process images. CNNs offer a wide variety of powerful object detection algorithms that assist in intelligent monitoring systems. In [1], using COCO Dataset and the YOLO algorithm, a system to detect social distance rule breakers is proposed. Pooranam, N., Sushma, Priya, Sruthi, S. and Sri, Dhanya [2] on similar lines proposed a system to measure the distance between two people in a public place using YOLOv4 and has an impressive accuracy of 90%. After detecting the people and calculating the distance between them, we ultimately distinguish those at risk at crowded places or public gatherings. This can help issue warnings to people or teach people to follow the proper protocol that can keep them and others safe.