The best facial recognition algorithms correctly matched individuals wearing masks 96 percent of the time in an evaluation last fall, demonstrating that people can potentially keep their masks on while having their photo identity automatically checked amid a pandemic, the Department of Homeland Security Science and Technology (S&T) Directorate said last week.
Without masks, the testing showed that the best systems correctly identified people 100 percent of the time, S&T said.
The evaluations were held over 10 days as part of the 2020 Biometric Technology Rally at S&T’s Maryland Test Facility.
“This isn’t a perfect 100 percent solution but it may reduce risks for many travelers, as well as the frontline staff working in airports, who no longer have to ask all travelers to remove masks,” Arun Vemury, director of S&T’s Biometric and Identity Technology Center, said in a statement.
Overall, the median performance of the systems tested with people wearing masks demonstrated a 77 percent identification rate and the median performance of the systems tested against individuals without masks was 93 percent, demonstrating that “performance can vary greatly between systems,” S&T said.
The testing involved 582 volunteers representing 60 countries. S&T tested six systems for face capture and 13 systems to match faces. The vendors and their systems are not identified in the initial test results.
Some of the systems performed poorly. S&T said that in face recognition testing of people without mask, the worst systems only identified 11 percent of the test subjects. For people with masks, the worst systems only identified 4 percent of the individuals.
The rally also tested one commercial iris recognition system, which was able to take a photo for 80 percent of people not wearing mask and only 67 percent of people wearing masks.