For the latest news and information about the coronavirus pandemic, visit the WHO website.
It turns out that face masks not only effectively prevent the spread of airborne diseases COVID-19 - Researchers have also been successful in blocking facial recognition algorithms.
In a report released on Monday, the U.S. National Institute of Standards and Technology found that face masks thwart even the most advanced face detection algorithms. The error rates varied between 5% and 50% depending on the capabilities of an algorithm.
These results are worrying for the face recognition industry, which has sought to develop algorithms that can be used to identify people through eyes and nose only when people turn face masks in the middle of the coronavirus pandemic.
Face masks are important tools to limit the spread of the disease, and U.S. governments require people to wear blankets. The masks caused problems with the facial recognition software and caused the technology companies to adapt. For example, Apple has released an update so that facial recognition works even when people wear protective covers.
Follow the coronavirus pandemic.
Face detection algorithms are based on transferring as many data points as possible to a person's image, and face masks tend to take away a lot of valuable identifying information. The algorithms are already so picky that improper lighting or a bad angle can deceive the technology, and masks make things worse, the study says.
The study found that an algorithm with an error rate increased from 0.3% to 5% when images of masks were displayed. The study tested the effectiveness of 89 face recognition algorithms against face masks.
The test examined the "one-to-one" fitting functions of the algorithms. Basically, a photo of a person was compared to another picture, but with a mask. NIST used 6 million images for its research and used masks digitally with different cover variations.
The study also found that the more that was covered by the nose, the more likely the mask was to interfere with the algorithms. Black masks were more likely to deceive the algorithms than blue masks, as research showed.
NIST said this was the first in a series of tests related to facial recognition and face masks. The agency plans to test algorithms specifically designed for coverage later this summer.
"With the outbreak of the pandemic, we need to understand how facial recognition technology deals with masked faces," said Mei Ngan, a NIST researcher behind the report. "We initially focused on how an algorithm developed before people with facial masks could affect the pandemic."
According to Ngan, NIST expects algorithms to improve the recognition of people with face masks.
Facial recognition researchers have compiled photos of people with face masks as data from which their algorithms can learn - in some cases without people's knowledge.
The NIST study reportedly used photos of people applying for immigration benefits and digitally altered mask photos of travelers entering the United States.
