Social Distance Measuring Based on Monocular Vision

Socia1 distance has been a growing concern since the COVID-19 pandemic broke out globally. Statistics indicate that keeping social distance is of great practical significance in slowing the spread of the pandemic. Traditional ranging methods rely on ultrasonic, infrared, laser or others. Unfortunately, most of these methods require Bluetooth modules or particular measuring sensors and need to fix hardwire devices on objects, which makes it costly and difficult to apply for measuring distances in various scenes. In order to reduce cost and extend application scope, this paper studies a novel ranging method based on monocular vision, which is proposed to estimate the distance between people in surveillance images. Our approach is to measure the social distance via the world coordinate relationship transformation or the principle of pinhole imaging after performing pedestrian detection. It is worth mentioning that this method only needs computer monocular vision technology, which is low in cost and suitable for an abundance of application scenarios. Through the experiment and analysis, our method shows good performance of social distance measuring in application.