Facial recognition up to 30% more accurate now than in 2010: NIST report

Facial recognition algorithms are up to 30 percent more accurate than they were in 2010, according to a recent NIST report.

The report by Patrick Grother and Mei Ngan, Performance of Face Identification Algorithms, includes results from algorithms submitted by 16 organizations. Performance was defined by recognition accuracy, as well as the time the algorithms took to match one photo against massive photo data sets.

“We studied the one-to-many identification because it is the largest market for face recognition technology,” Grother said. “These algorithms are used around the world to detect duplicates in databases, fraudulent applications for passports and driving licenses, in token-less access control, surveillance, social media tagging, lookalike discovery and criminal investigations.”

Four research groups enrolled in both the 2013 and 2010 tests, which gave NIST the ability to compare performance improvements over time. NIST found these groups had improved their performance on the tests by from 10 and 30 percent. One unnamed organization decreased its error rate from 8.9 percent in 2010 to 6.4 percent in 2013.

The database of images used to perform these tests included 1.6 million faces. Some were ANSI/NIST ITL 1-2011 Type 10 standard, others were ISO/IEC standard and 140,000 of them were images taken on webcams in uncontrolled environments.

The tests unsurprisingly performed best on the high-quality ISO standardized images, and no algorithms worked well with the webcam images.

According to NIST, the study also shows that rates of missing facial matches increase as the database size increases as expected, but that it does so only slowly. When the number of facial images increased by a factor of 10—from 160,000 to 1.6 million—the error rate only increased by about 1.2 times. This slower-then-expected growth in error rates occurs in many natural phenomenon, and “is largely responsible for the operational utility of face identification algorithms,” explains Grother.

Images of older individuals were identified more accurately than those of younger persons, suggesting that we become steadily easier to recognize using facial recognition software, and more distinguishable from our contemporaries, as we age.

Reported previously in BiometricUpdate.com, a recent study published in the journal of Psychological Science by researchers at the University of Texas at Dallas and the National Institute of Standards and Technologies described a series of experiments that demonstrates there is potentially more information for biometrics-based identity recognition in images of people than the face alone.

Article Topics

accuracy  |  algorithms  |  face  |  facial recognition  |  NIST  |  test