How Computer Vision Helps in Monitoring and Measuring Social Distancing


As the impact of CVID-19 is receding in many countries, a number of them are preparing to ease the restrictions. However, along with the newfound freedom comes to the threat of people abusing it, particularly when an individual’s carelessness could put several others at risk. Consequently, enforcing strict social distancing measures is more critical now to avoid a second wave of suffering. 

As straightforward and socially responsible as it might sound, not every individual is keen to wear masks or maintain a social distance. Neither is it manually possible for officers to enforce strict decorum in public places. Several locations such as the Miami beach Park had to close after reopening its premises because the crowds defied the social protocols. 

For authorities, the biggest challenge now is to ensure that everyone is following protective measures. And predictably, it is no easy task. 

Fortunately, thanks to computer vision technology, there is a means to enforce radical measures to flatten the curve. 

Social Distance Detection using OpenCV

Any tracking system has a well-defined role to observe the moving objects (like people in the streets). In addition, the object tracking software can predict the direction of motion.  

OpenCV is the most popular open-source machine vision library that includes more than 2500 algorithms. 

Two students in India have endeavored to develop a code that will enable detection on individuals. This Single Shot object Detection done through MobileNet and OpenCV is available for everyone. 

The program will display a bounding box on every person that is detected through the system. The software will also display the distance of the person from the camera. If the distance between the two people is less than 2 meters, the green bounding box will turn to red. 

Zensors to do Live Tracking 

Zensors technology, invented by a team at Carnegie Mellon University, is currently working with governmental authorities in a few countries such as the US and India to enforce social distancing. 

The technology combines computer vision and analysis data from existing CCTV cameras to help track the crowd in real-time. Initially intended for commercial purposes, now the technology has been repurposed to cater to concerns of COVID-19. Zensor can now provide information on how many people have assembled in an area and whether they are following social distancing. Moreover, it will also tell you when the surface was last cleaned and whether further cleaning measures have to be expedited. 

AI-enabled Contactless Screening 

Contactless thermal screening enables for speedy clearance of passengers at checkpoints while maintaining a safe distance. Megvii’s latest Koala Solution has already been installed in several supermarkets and other public places across Beijing. 

This AI-enabled CV solution blends body and face detection along with dual sensing, using infrared cameras and visible light. It can improve the accuracy of temperature screening without requiring any physical contact. Higher body temperature will trigger a warning system that will allow a staff member to approach the individual for manual screening. 

In India, Arvi, a health-tech startup, has launched kiosks with contactless thermal scanning. It used Ai technology integrated with deep learning and facial recognition to detect those with high temperatures. Systems like this are in place in public facilities, including markets and railway stations. 

Tracking is a field with immense applications of Computer Vision. Today, when the world needs it the most, employing computer vision can help assess public health risks and plan to reopen with sufficient precautions.

Telecomdrive Bureau
TelecomDrive is an effort to create a unique content focused platform for the telecoms and communications segment.