Wearing a face mask makes it harder for facial recognition algorithms to see you, according to a new government study

Wearing a face mask makes it harder for facial recognition algorithms to see you, according to a new government study
  • A new government study found that wearing face masks makes it harder for facial recognition algorithms to recognize people.
  • The typical error rate for most facial recognition algorithms is 0.3%. When subjects in the study wore a face mask, the error rate rose to anywhere from 5% to 50%.
  • The algorithms are less likely to recognize someone wearing a mask over their nose, the study found. Wide masks that cover the full face were more likely to trick facial recognition than round masks.

Wearing a mask doesn't just limit the spread of COVID-19 — it may also makes it harder for facial recognition algorithms to identify you, according to a new study.

The National Institute of Standards and Technology tested 89 facial recognition algorithms, which have an error rate of around 0.3%. But when those algorithms were used on subjects wearing face masks, the error rate jumped to anywhere from 5% to 50%, researchers found.

The federal government conducted the study in part because it's using facial recognition algorithms to track people's movements and identify suspects across the US, where mask-wearing is now recommended to slow the spread of COVID-19. The NIST study was carried out "in collaboration with" the Department of Homeland Security and US Customs and Border Protection, both of which use facial recognition technology.

Complimentary Tech Event
Discover the future of SaaS in India
The 6-part video series will capture the vision of Indian SaaS leaders and highlight the potential for the sector in the decades to come.Watch on Demand
Our Speakers
Girish Mathrubootham
Brian E. Taptich

The study, which was earlier reported by The Verge, suggests that wearing a face mask would make an individual less likely to be identified by facial recognition. But not all masks are created equal — people are less identifiable when wearing a mask over their nose than when they leave their nose uncovered, the study found.

Findings also suggest that dark masks are better at evading facial recognition than blue surgical masks, and that wide masks that cover a person's entire face beat the algorithms more than round, N95-style masks.


"None of these algorithms were designed to handle face masks," said NIST computer scientist Mei Ngan, who authored the report. "With respect to accuracy with face masks, we expect the technology to continue to improve."

Ngan added that NIST plans to test newer algorithms later this summer that were built with face masks in mind.

Kurt Opsahl, deputy executive director of the privacy-focused Electronic Frontier Foundation, said on Twitter that even though the study was conducted by the federal government, its findings could help citizens who want to evade surveillance.

Read the full findings of the NIST study here.