NEC’s face recognition technology tops NIST testing again

July 16, 2014 - 

NEC Corporation announced that, for the third consecutive time, its face recognition technology was ranked the highest in performance in NIST’s Face Recognition Vendor Test 2014 performed by the U.S. National Institute of Standards and Technology.

The NIST report is considered an industry benchmark for assessing how accurate a face recognition software solution is in one-to-many searching in various circumstances.

This marks the company’s third consecutive time its technology has performed better than other facial recognition solutions, following the 2009 Multiple Biometric Grand Challenge and 2010-2011 Multiple Biometrics Evaluation. The most recent Face Recognition Vendor Test (FRVT) test for one-to-many searching (class C) assessed the accuracy of database enrollments and personal identity searches using face images of over 1.6 million people, including “high quality” criminal mugshots, “low quality” webcam snap shots, and “standard quality” passport/visa photo images.

The FRVT evaluation, which saw participation from 12 commercial vendors and four academic institutions, comprised of many different test situations designed to assess how accurate each solution was based on their measured error rates.

The test concluded that NEC’s search accuracy outperformed the other vendors by a long shot, as well as greatly improved from the previous MBE 2010 testing. When it came to the test of rank one miss rates for recognition in a population size of 1.6 million, NEC saw a rate of 4.1%, successfully outscoring the second best vendor by a wide margin. And NEC’s face recognition algorithms performed more rapidly than the other vendors, at a rate of 0.2 seconds per image to generate and search an image against 1.6 million records.

“These results are extraordinary when considering the prior NIST MBGC 2009 and MBE 2010 testing in which NEC also registered the top ranking and superior performance,” said Raffie Beroukhim, vice president of NECAM’s biometrics solutions division. “These improvements demonstrate that NEC is not only outperforming our competitors, but also advancing our own solution. NEC’s face recognition matching algorithm improved the most in one-to-many search accuracy as compared to the other industry participants.”

The NEC automated face recognition solution is being used more frequently in industries other than law enforcement, including retail and hospitality.

The global facial recognition market continues to see explosive growth. A recent ReportStack report projected that the market would see a 26.6 percent compound annual growth rate from 2013 to 2018.

Leave a Comment

comments

About Stephen Mayhew

Stephen Mayhew is the publisher and co-founder of Biometrics Research Group, Inc.. His experience includes a mix of entrepreneurship, brand development and publishing. Stephen attended Carleton University and lives in Toronto, Canada. Connect with Stephen on LinkindIn.