Design and Development of 3D Social Distancing Detection Module

Ng, Yong Ting (2021) Design and Development of 3D Social Distancing Detection Module. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text NG YONG TING_Full text.pdf Restricted to Registered users only Download (3MB)

Abstract

Rampant coronavirus disease 2019 (COVID-19) brought a global crisis, with its deadly spread to more than 180 countries, with about 3,519,901 confirmed cases and 247,630 deaths worldwide as of May 4,2020. Lack of any active therapeutic agent and lack of immunity to COVID19 increases the vulnerability of the population. Motivated by this notion, this project aims to design and develop a real-time 3D social distance detection module to ease the authority departments to do monitoring at community park, by using a fusion algorithm of 2D computer vision and 1D depth detection to detect social distance. Computer vision for two-dimensional images (X&Y dimensions) has the limitations of detecting scene depth (Z dimensions), but it is able to detect multiple faces of images captured in the camera. Therefore, when computer vision detects the human body, lidar sensors are used to determine the depth of the human body Z-dimensional. An algorithm to calculate the 3D distance will be form by using the fusion of image processing technology and LiDAR technology. To achieve this objective, an experiment real-time video of people walking down the street was used to test the proposed process. The results demonstrate that the suggested approach will evaluate social distancing measures between different individuals in real time. However, the distance between the human and the LiDAR sensor must be between 0.3 and 12.0 meters, which was the project's main limitation. Nonetheless, this project has contributed to society a fusion algorithm that can efficiently monitor social distancing between individuals and can be modified for use in other environments such as the office, restaurant, and classroom.