Visualizing Social Distancing: a Computer Vision Approach

Chigozie Ofodike, Stacy Fortes, Dr. Juan Li and Dr. Daehan Kwak Kean University, School of Computer Science and Technology, Union, NJ 07083

When the COVID-19 pandemic began in early 2020, it was widely considered a passing matter—one that would only stay for a short time. After nearly a year, this virus has been spreading relentlessly and perpetually. The question has shifted from “When will COVID-19 end?” to “How do we live with COVID-19?” as people wait for the leading scientists and pharmaceutical companies to do their due diligence. To mitigate the spread, individuals have been practicing social distancing and wearing masks. While some wait for a safe and effective vaccine, others grow restless in quarantine, wanting to go out, spend time with friends and have social gatherings. Temptation to break social distancing protocols sometimes becomes too strong to bear, despite the fact that it is essential in slowing the spread. Health professionals have measured social distancing on a macro-level to study the efficacy of social distancing to prevent COVID-19 transmission. However, such studies are insufficient due to their imprecision of data sourcing and processing.

In this study, solutions to measure social distancing on a micro-level are proposed in order to cover this lapse. A system is being built using machine learning and computer vision to show the percentage of people social distancing based on images or videos. The system is implemented by using Python’s Computer Vision libraries that allow for motion and object detection. For each frame of an image or video, a feature extraction algorithm is introduced. Then, applied statistics is used to determine the percentage of people social distancing in any given environment. This application will hopefully act as a method to reinforce the need for social distancing especially in the circumstance of a major pandemic, and could support healthcare professionals to study and explain the impact of social distancing on the spread of COVID.

Additional Abstract Information

Presenters: Chigozie Ofodike, Stacy Fortes

Institution: Kean University

Type: Poster

Subject: Computer Science

Status: Approved

Time and Location

Session: Poster 5
Date/Time: Tue 12:30pm-1:30pm
Session Number: 4021