NGHIÊN CỨU GIẢI THUẬT PHÁT HIỆN KHẨU TRANG CHO ROBOT TRONG PHÒNG CHỐNG BỆNH DỊCH

Abstract

The mask detection feature has significantly advanced in image processing and computer vision, making a crucial contribution to combating the COVID-19 pandemic. This paper presents a method utilizing the Caffe model in combination with the MobileNetV2 architecture, designed to apply to embedded devices with limited computational capabilities such as Raspberry Pi, NVIDIA Jetson Nano, etc. This method enables real-time mask detection even when individuals are wearing hats or glasses. Experimental results were obtained using a custom-built dataset from students of the Faculty of Electrical and Electronics Technology at the Ho Chi Minh City University of Industry and Trade, achieving an accuracy rate of up to 93%.