Deep Learning-Based Smart Mask for Social Distancing

ABSTRACT

The COVID-19 is having a significant impact on our everyday lives. One of the measures put in place to slow the spread of the disease is social distancing. Because of the abrupt implementation of these laws, it become impossible for a person to follow this form of lifestyle. So we have developed a device that can assist both normal and visually disabled people in maintaining physical distance from other people by using deep learning to identify an individual and verify if the other person is six feet away from the user so that social distance can be sustained, and virus spread can be reduced. Our architecture includes a Webcam and a Raspberry Pi 4 model B to run an advanced image processing algorithm SSD MobileNet for human detection. SSD MobileNet is implemented to determine if the other individual is physically close to the user. If an individual does not adhere to social distancing, an automated reading aid is activated, and an audio warning is sent to the user. Since the whole setup is light and portable, it can be conveniently installed with any mask. The job is evaluated in real time. The results show that the proposed device improves mobility, convenience, and ease of navigation for both normal and visually disabled people in this pandemic situation. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.