The National Institute of Standards and Technology has published a report that points to rapid advancements in the U.S. marketplace for facial recognition software platforms.
NIST said Friday it evaluated 127 software algorithms from 39 different developers and found that face-based biometric matching algorithms showed 20 times better searching performance over the past four years.
Current software products only failed in 0.2 percent of searches this year compared with similar tests in 2010 and 2014 when facial recognition tools failed 5 percent and 4 percent to match subjects, respectively.
Patrick Grother, a NIST computer scientist and an author of the new report, said the decline in error rates indicates new algorithms have revolutionized the field.
“The test shows a wholesale uptake by the industry of convolutional neural networks, which didn’t exist five years ago. About 25 developers have algorithms that outperform the most accurate one we reported in 2014,” Grother added.