Social Distancing Monitoring System Based on Image Processing

Abstract

This project is about using Raspberry Pi to create the monitoring system for the detection of social distancing status for queueing up situations in public places. The objective of conducting this study is to develop a social distancing monitoring system based on Image Processing, calculate the distance of 1 meter based on the image pixel of queueing up people, and to compare the best position/angle of the camera location to get a good 1-meter accuracy. The scopes of the project are the system only focuses on  queueing up situation, the data accuracy is limited to the experiment setup and Raspberry Pi Camera Board use, and the subject of people queueing up is pre-defined in this project. Raspberry Pi associated with its camera module is used for the monitoring purpose. OpenCV and Python is implemented to develop the program associated with Caffe models for people detection. The best angle for accurate detection is from side view and the system can run on real-time when tested over 10 runs. The main suggestions to improve the overall system is by using a camera with higher resolution and adding more samples for the detection.