After analyzing how the Covid-19 virus is spread, scientists found that microdroplets play a crucial role in spreading the virus. Hence, social distancing is essential to a distance of at least 1.5 meters.
Scientists at EPFL’s Visual Intelligence for Transportation (VITA) Laboratory have previously developed an algorithm to detect the presence of another car or a pedestrian on the road and instruct the self-driving car to slow down, stop, change direction or accelerate.
Now, scientists tweaked this algorithm and developed a 3D detector called MonoLoco, to detect whether individuals maintain the right distance to prevent infection. The detector can be easily attached to any camera or video recorder, or smartphone.
The detector calculates the dimensions of human silhouettes and the distance between them. It also identifies people’s body orientation, determining how a group of people interact – and especially whether they’re talking – and evaluate whether they’re staying 1.5 m apart. This is possible because it uses a different calculation method than existing detectors.
More importantly, the detector keeps the faces and silhouettes of people completely anonymous. It only measures the distances between their joints.
After capturing the picture or video of a given area, it converts the people’s bodies into unidentifiable silhouettes sketched out with lines and dots. That’s how it maintains people’s privacy.
Lorenzo Bertoni, a Ph.D. student at EPFL’s Visual Intelligence for Transportation (VITA) Laboratory, said, “We came up with several possible applications for our program during a pandemic. On public transport, of course, but also in shops, restaurants, offices and train stations – and even in factories, since it could let people work safely by maintaining the necessary distance.”
The source code of the algorithm is available on the VITA website.
- Lorenzo Bertoni, Sven Kreiss, and Alexandre Alahi, “Perceiving Humans: From Monocular 3D Localization to Social Distancing,” IEEE Transactions on Intelligent Transportation Systems, 2021. DOI: 10.1109/TITS.2021.3069376