Human Detection, Counting and Social Distancing Measurement System

Ong, May Lin (2022) Human Detection, Counting and Social Distancing Measurement System. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Ong May Lin.pdf Restricted to Registered users only Download (4MB)

Abstract

People counting and social distancing measurement are a crucial and challenging problem in visual surveillance. Also, automatic monitoring of the number of people and social distancing measurement in public areas is also important for safety control and urban planning especially at pandemic period such as Covid-19. In this proposal, human detection, counting and social distancing measurement system is proposed. This is an efficient on counting-approach and distancing measurement that easy to implement for real time monitoring systems. An indoor video frame was used as input, and the YOLOv3 algorithm was employed for human detection. The transfer learning methodology is also implemented to increase the accuracy of the model on overhead view. The accuracy achieved by model with transfer learning is 97.75%. Besides, people are counted based on the bounding box and Euclidean distance is for social distancing measurement. A violation threshold is established to evaluate whether or not the distance value breaches the minimum social distance threshold. Therefore, through this project the number of people and social distancing measurement in the shop or shopping mall can be easily known to avoid number of people overload and keep social distance.