부산대학교 도서관

Since the massive spread of the COVID-19 epidemic in 2020 and the recent spread of influenza, the contactless delivery trolley has met the demands of the government for logistics against epidemics and the needs of logistics companies for deliveries. This paper designs a recognition method based on YOLOv5 (You Only Look Once version 5) for detecting room numbers and QR codes for the needs of anti-epidemic delivery trolleys and implements tasks such as stopping the anti-epidemic trolley at a specified room number and opening different warehouses according to different QR code information. The experiments improved the recognition by filtering the pepper noise through mean filtering. Experimental results show an increase of 2.105 % in the recognition rate of pre-processed QR codes, with an average accuracy rate of 98.9%. The delivery trolley also achieves a stable recognition rate of around 98.81% for room numbers.