Remote identity verification technology provider Netki announced the release of newly updated face matching technology for its OnboardID system at the North American Bitcoin Conference this week.
The update to the tech aims to improve the system’s ability to determine whether faces scanned from various sources belong to the individual in question.
“Our face matching algorithm now outperforms industry-leading solutions by more than 60%,” said Steve Scott, Director of Software Engineering at Netki. “We’re especially proud of the algorithm’s ability to successfully match a wide variety of skin tones, from very light to very dark, as well as find and match faces in darker lighting conditions,” he added.
OnboardID is used by clients — typically financial institutions — to validate the identities of both new users and existing customers. It works by directing users to take a selfie along with a picture of their valid photo ID, and then uses face matching technology to determine whether the two pictures submitted match.
“In the past, we licensed best-of-breed third party algorithms inside of our face matching platform, but these solutions often were unable to successfully find and match faces in the wide variety of environments and lighting conditions end users experience in the wild,” said Justin Newton, CEO of Netki. “Frankly, we were ecstatic to see such a significant increase in face detection and matching with our newest update.”
With increased accuracy there is also potential for the reduction in costs to users, as there are likely to be fewer cases that require manual inspection of pictures that don’t pass the face matching test.
“Manual inspections are costly and time-consuming for customers. With the most recent update to our in-house technology, we have been able to reduce the number of cases requiring manual inspection by 83%!” said Newton. “It’s a dramatic improvement, which will reduce costs for our customers as well as speed their new account creation.”
Netki is also placing emphasis on its improved system’s ability to identify people across a variety of ethnicities and skin tones, addressing what is a major issue in face recognition technology today in light of a recent NIST study that showed the majority of systems used today carry with them racial bias that can discriminate against visible minorities and women.
“We wanted to ensure that Netki’s algorithms were different than the others,” said Dawn Newton, COO of Netki. “That focus enabled us to develop our latest facial recognition algorithm, that recognizes faces from around the globe.”
A rollout of Netki’s update to their face-matching tech is happening now and is expected to reach all of its customers over the next few weeks.