Artificial intelligence and Covid-19: the challenge of visual social distancing
22 July 2020 | Written by La redazione
A study carried out by a collaboration between the Italian Institute of Technology (IIT) and the University of Verona published in the scientific journal "IEEE Access"
Maintaining physical distance has proven to be one of the most effective measures to contain the Covid-19 pandemic. Monitoring compliance with this precautionary measure is becoming fundamental to limiting the mobility of individuals as little as possible without however exposing society to further infections. In this context, a new challenge for artificial intelligence is visual social distancing, i.e. the automatic measurement, starting from an image acquired by a video camera or camera, of interpersonal distance with a characterization of the social relationship. The problem of visual social distancing is at the heart of the study conducted by a team of researchers from the Italian Institute of Technology, the University of Verona and the University of Glasgow, together with Humatics, a spin-off of the Scala University specialized in systems machine learning.
The research “The visual social distancing problem”, published in the scientific journal “IEEE Access”, for the first time traces the path that research in the field of computer vision must follow, to arrive at tools capable of supporting the human being in the task of monitoring compliance with the spacing rules.
Developing technologies that can support us in understanding the dynamics related to the creation of gatherings or the movements of people in public and private spaces, to modify them ad hoc and allow the free movement of citizens, minimizing health risks, has become crucial. In this publication, the problem of visual social distancing is introduced for the first time, as a fundamental measure for non-invasive and respectful privacy monitoring for the purpose of maintaining social distancing, and for the statistical evaluation of the risk level of specific areas, such as for example, boarding areas of airports, supermarket checkouts, ATMs. The research efforts will thus be directed towards the creation of systems that are able to understand the social context and the specific situation, and provide an effective tool for maintaining spacing measures capable of identifying only situations of potential danger without creating fakes alarms and respecting citizens’ privacy.
A distance for every need. “Our study highlights that it is not enough to create a system that measures the geometric distance between people – explains Alessio Del Bue, a researcher at the Italian Institute of Technology and author of the study – it is essential that monitoring technology is able to understand the social context and produce alerts only in case of real danger “.
“The distance between people could vary with parameters such as the degree of kinship, the use of individual protection devices – such as masks, screens or other – but also in the cases of the elderly or people with disabilities who need assistance” says Marco Cristani, Professor of Information Processing Systems at the Computer Science Department of the University of Verona and co-founder of the spin-off Humatics, another author of the study.
Although they have a fairly large taxonomy relating to the different types of gatherings, the artificial intelligence systems used to date for monitoring from social distance are still approximate when the scenario to be monitored is composed of numerous people with different social relationships.
By addressing the technological problem identified for the first time since this publication, in the future it will be possible to have technologies capable of monitoring social distancing effectively and non-invasively, but also robots and systems based on augmented reality or mixed reality more performing, in addition to new tools for the study of all those human pathologies that influence the physical and social distance between individuals, such as depressions and behavioral disorders.