CrowdSurge: A Crowd Density Monitoring Solution Using Smart Video Surveillance with Security Vulnerability Assessment

Abstract

—Overcrowding and crowd density monitoring in various places and establishments are being implemented since the pandemic, which helps observe social distancing. This study is about the development of a crowd density solution by utilizing YOLOv4 and Closed-Circuit Television (CCTV) called CrowdSurge. The practice of CCTV has been around for so many years with proven benefits. This has been combined with the state-of-the-art YOLOv4 algorithm that provides high video analytics and object detection performance. With the combination of the said technology and algorithm, it will serve as a smart surveillance system. A system and mobile application have been developed, and the YOLOv4 deep learning detection model was used to detect various set of scenarios considered to assess if the model executes according to the actions assigned in the experimental set-up. The browser-based application was tested using CVSS or Common Vulnerability Scoring system, which shows that the severity level of most vulnerabilities is low and has a minor impact on the system. Based on the overall usability testing and statistical results, the respondents are satisfied with both surveillance system and mobile applications developed in terms of functionality, usefulness, and aesthetics. Therefore, using the developed system in real-time surveillance can aid in crowd density reduction in an area.