IMPERSONAL: An IoT-Aided Computer Vision Framework for Social Distancing for Health Safety

Abstract

Recently, pushed by the COVID-19 pandemic, the need of respecting social distancing has motivated several researchers to define novel technological solutions to monitor and track user movements. Information and Communications Technologies (ICT) world has addressed this challenge by means of the use of different technologies, such as Bluetooth, in order to track user interdistance and encounter time. Technology solutions should be able to not only track contacts, but also alert users to restore social distancing. In this article, we present IMPERSONAL framework, with the twofold aim of both: 1) tracking and monitoring social distancing and 2) alerting users in case of gatherings. The framework is based on a subnetwork of computer vision-based devices that are adopted to monitor and track users’ movements to estimate their interdistance and compute the encounter time. Such information is then the input to an Internet of Things subnetwork, aiming to retrieve the anonymous IDs of people belonging to a gathering, as well as to send alert messages to them. We assess IMPERSONAL framework by means of extensive Monte Carlo simulations and experimental results, showing its effectiveness in terms of accuracy in correctly identifying users and gatherings in videos taken from live cameras, both in case of indoor and outdoor real scenarios. The benefits of the IMPERSONAL framework are expressed in terms of the ability to track people, solve gatherings, and send warning messages.