NIST Study Finds Masks Undermine Face Recognition Accuracy
July 30, 2020
A study conducted by the National Institute of Standards and Technology showed that face masks undermine the accuracy of facial recognition algorithms. The NIST study tested digitally applied masks of various shapes on 89 commercial algorithms. The result were error rates between 5% and 50%. The algorithms tested were all created pre-Covid-19. NIST plans to test facial algorithms developed with face masks in mind later this summer. A previous NIST study released at the end of last year found that false positives are up to 100 times more likely for Asian and African American faces when compared to White faces. EPIC has previously launched a Ban Face Surveillance campaign and called for a facial recognition moratorium across the globe, as well as suspension across the federal government and in U.S. schools.