Face masks effective in stopping the spread of COVID-19, but many people believe they can break facial recognition algorithms. Now, facial recognition companies are reworking algorithms to identify individuals on features visible even with masks — the portion of the face above the nose.
With one’s nose, mouth, and cheeks covered up, it’s not hard to believe that identification would become more difficult. Typically, facial recognition works by comparing measurements between different facial features in one image to those in another.
The US National Institue of Standards and Technology (NIST) conducted a study on masks’ effects on facial recognition algorithms. They found that face masks caused the error rate of widely used algorithms to spike to between 5 to 50 percent. The more of the nose that was covvered, the more difficult it was for algorithms to identify the face.
As a result, companies are racing to have artificial intelligence rely on other features, like the eye region, for facial recognition, CNN reports.
Masks and Facial Recognition
What will that mean? Mask wearers are more likely to be recognized despite covering most of their faces with coverings. However, if people were to wear both masks and sunglasses, the updated algorithms wouldn’t have much to work on.
However, as companies shift algorithms to better identify masked faces, the risk of misidentification rises. Clare Garvie, a researcher at Georgetown Law Center on Privacy and Technology, spoke to CNN about facial recognition relying on less information to identify individuals
“I just see the misidentification risk vastly increases because they have less biometric to go off of,” she told CNN. “It doesn’t matter how they tune the algorithm.”
In instances where facial recognition is used by law enforcement during investigations, they may misidentify a suspect. The Verge reports that the NIST is planning another test on masks’ effects on facial recognition. The new test will be target algorithms tuned for mask wearers.
Mei Ngan, an author of the NIST report and computer scientist, told The Verge, “With respect to accuracy with face masks, we expect the technology to continue to improve.”
Facial Recognition Software Creating Racial Bias in Law Enforcement?
California Close to Banning Facial Recognition Tech on Police Body Cameras
What You Need to Know About Facial Recognition at Airports