Studies Show Masks Can Wreck Facial Recognition Algorithms

At Allconnect, we work to present quality information with editorial integrity. While this post may contain offers from our partners, our opinions are our own. Here’s how we make money.

It’s no secret that face masks are now a necessary part of our everyday lives. However, the mandatory masks we’re wearing for public safety aren’t necessarily meshing well with the facial recognition algorithms powering the devices that are also part of our daily routines.

A new study by the National Institute of Standards and Technology (NIST) shows wearing face masks in the proper fashion (covering the nose and mouth) causes the error rate of commonly-used facial recognition algorithms to spike between 5%-50%.

“We have begun by focusing on how an algorithm developed before the pandemic might be affected by subjects wearing face masks. Later this summer, we plan to test the accuracy of algorithms that were intentionally developed with masked faces in mind,” says computer scientist at NIST, Mei Ngan.

Black masks were found to cause more error than blue and devices were less likely to recognize a face when the nose was also covered. Officials are instructing the public to cover both the nose and mouth to avoid risk of infection.

The report only included one type of facial recognition called one-to-one matching, often used in border and passport control situations. The method used in mass surveillance situations, known as one-to-many, was not included in the report but are generally thought to be more error-prone.

If face masks are disabling one-to-one matching, they’re most definitely disabling one-to-many recognition as well due to the difficulty of distinguishing faces in a crowd. Now, industry leaders are looking to improve facial recognition technology to identify users only by the area around their eyes.

NIST has announced plans to continue studies on masked users along with one-to-many recognition systems.

Bookmark our Resource Center for more tech updates, studies and expert looks at the ways we connect with our communities and the world.

Written by:

Taylor Gadsden

Writer, Broadband & Wireless Content

Taylor is a veteran member of the Allconnect content team and has spearheaded a number of projects, including a data piece on the top fiber cities in the U.S. and a troubleshooting guide … Read more