Auto-SDist: An Automatic COVID-19 Social Distancing and Room Management System Architecture for Indoors

• Main Authors: Shutt, Kenyon, Wan Nawawi, Wan Fatimah, Schulte, Gillian, Martinez-Sainz, Juan, Ufuktepe, Ekincan, Palaniappan, Kannappan
• Format: Conference Proceeding
• Language: English
• Subjects: Cameras; COVID-19; Human factors; Object detection; Pandemics; Protocols; Systems architecture
• ISSN: 2332-5615
• DOI: 10.1109/AIPR52630.2021.9762165

The COVID-19 pandemic has changed the world in unprecedented ways. Due to its high levels of transmissibility and fatality rates, the Centers for Disease Control and Prevention (CDC) continues to update their recommendations about safe human interaction. However, enforcing these standards using manual labor can be expensive and subjective. Thereby, this study aims to help public establishments enforce CDC protocols more effectively. We propose an automatic computer vision-based enforcement and monitoring system for room capacity and social distancing standards, which is assembled from cheap, accessible, off-the-shelf components. The proposed system, Auto-SDist, consists of several modular units, which can be installed into any given room as shown in Fig. 1. Live camera output is fed to a graphics processor, where an object detection algorithm detects the people in the room, then the system counts, and checks distances between the detected people. This occupancy and distancing information is then transmitted from the camera unit to the controller unit, which will then trigger actions in the various output devices. Each of these modules are connected via standard RJ45 cables, which carry both power and communications.