Mobile Patrol Robot for Temperature Measurement and Mask Detection using Deep Learning

Abstract

Wearing a mask especially in a crowded place is a must now days in order to prevent from the Covid-19 virus to spread wider but people tend to take lightly about wearing mask. The objective of this study is to develop a face mask detector and temperature sensor patrol robot which can moves on its own. Raspberry Pi 3b+ was used as the main component for mask detection and temperature measurement while Arduino Uno was used to control the motor movement based on input it receives from IR sensor. The mask detection system has achieved 58% to 73% accuracy when people wear a mask and 58% to 66% accuracy for the people who were not wearing a mask which was using a Tensorflow Lite algorithm of pre-trained Mobilenet SSD model. The pre-trained model has high accuracy for not wearing mask detection compares to wearing mask detection and also have a delay of 4 seconds for every video frame inputs. This study can be improved by using a Raspberry Pi 4 board which have higher computational power that can run the pre-trained model much faster and less delay.