Face masks not only help prevent the spread of the novel coronavirus but they’re also proving to be a bit of a nuisance for facial recognition technology.
In a report published by the US National Institute of Standards and Technology (NIST), it has been found that face masks are causing the facial recognition algorithm to fail as much as 50 percent of the time.
The researchers at NIST used about 6 million images, and applied face masks digitally, with different colours and variations. The study says that face masks seem to be thwarting even some of the most advanced facial recognition systems out there.
The Challenge for Facial Recognition Tech
During the research, the facial recognition software was asked to match the face of a person to one of the same person wearing a mask. The report says that the algorithm failed to make a correct match between 5 to 50 percent of the time.
One algorithm, that had a failure rate of 0.3 percent while scanning faces without masks, surged to 5 percent while analysing images with face masks.
It’s no surprise that technology is failing as it works by comparing the measurements of different facial features in one image to those in another.
Since face masks block almost half of the face, it’s difficult for the algorithm to interpret the data accurately as there is not much information available.
How to Get Around This Problem?
The current scenario has become a challenge for the facial recognition industry as its scrambling for ways to find a fix around this problem.
Most companies are deploying algorithms that can identify people through their eyes and nose alone.
Even Apple had recently pushed an update for the iPhone, where its Face ID unlocks the smartphone even if a user is wearing a face mask. Other places facial recognition technology is used is at security checkpoints at the airport and places with a high level of security.
According to a CNN Business report, some companies have said that they have facial recognition technology that can work with face masks and that the developers are working on algorithms to come up with a solution.
Some of the possibilities are training the algorithm with images of a lot of people wearing masks or by targeting specific sections of the face like the eyes, ears or upper part of the nose.
The Way Forward
Facial recognition technology is an essential part of the security infrastructure of a country. Many have criticised the abuse the technology to spy on people but one can’t help ignore its benefits.
NIST is expected to release a report in the coming days that will tell the accuracy of facial recognition tech that has been specifically created to read faces with masks on.
Meanwhile, commercial firms are enhancing their existing technology to identify people wearing masks.
A Japan Times report has said that there is a biometric authentication device that scans the irises of the person alone with the face and it’s been able to deliver results with an error rate of one in a billion.
There are possibilities that future facial recognition technology will merge a couple of biometric scanning solutions that will be able to bypass the problem of face masks.
(At The Quint, we are answerable only to our audience. Play an active role in shaping our journalism by becoming a member. Because the truth is worth it.)