NIST sees drastic improvement in facial biometrics accuracy as market grows

Biometric facial recognition technology has improved “spectacularly,” with error rates for the top products estimated by the biometrics testing project leader at the National Institute of Science and Technology (NIST) to be improved by a factor of 50 or 60 over the algorithms tested in 2010, AFCEA’s Signal reports.

“The algorithms now are spectacularly more successful [at matching one to one matching] than they were when we first tested this technology in 2010,” NIST’s Patrick Grother told an audience at AFCEA International’s Federal ID Forum and Expo.

The remaining errors are mostly due to the long-term effects of aging and injury, he says. Poor photo quality can also increase error rates from below one percent to above five percent.

Despite the significant advances, technologies tested by NIST show a wide range of performance.

“There’s a great deal of variation across the industry in terms of algorithm capability. It’s not commoditized technology, you can’t buy anyone’s … You have to do due diligence,” Grother cautions.
NIST plans to publish additional research on demographic dependencies, or disparate effectiveness for people based on differences in age, sex, or skin color.

Futurum Research Principle Analyst and Broadsuite Media Group CEO Daniel Newman considers the social and regulatory landscape around facial recognition in a Forbes editorial, observing that experts agree that oversight and security of biometric data are lacking. The potential, however, for customized customer experiences and fine-grained marketing data, is enormous, if precision and privacy challenges are conquered.

Transparency is the key to support wide-spread adoption, Newman argues.

“Before the technology is unleashed into the corporate world, companies need to fully understand it,” he says. “They also need to be able to explain to users exactly what data they are looking at, what information is being stored and how it’s being used.”

Implementations lag customer demand

While a majority of consumers want to use facial recognition (55 percent) in retail scenarios, only 20 percent of brands offer it, according to research from Elastic Path reported by Net Imperative. Similar numbers are also seen for how many shoppers want to use voice commerce (57 percent) compared to how many brands let them do so (23 percent).

The “The Sci-Fi Shopper: How to Future Proof Your Brand for the New Customer” white paper surveyed shoppers on four pain points in retail experiences, and found that the number of people who experience each pain point is about 20 percent higher than the number of companies who perceive each one as an issue.

Two-thirds of shoppers want checkout-less payments, a service only offered by 18 percent, and smart devices are desired by 58 percent of shoppers, but only used by a quarter of brands surveyed.

“When it comes to new technology, brands continue to chase shiny objects. Consumers just want to buy the products they need with as little effort as possible,” says Darin Archer, CMO of Elastic Path. “Companies need a strong commerce backend that supports the latest innovations without sacrificing customer experience. It’s up to brands to make any experience shoppable. With the right tech, they can enable consumers to purchase almost without thinking, anytime and anywhere.”

CB Insights suggested recently that facial recognition is about to play a major role in a wide range of activities.

Applications demonstrated at events

Attendees of the Go West 2020 conference for event professionals will be able to experience a range of facial recognition technologies through a partnership with Zenus, Event Industry News reports.

“Use of face recognition in events is here,” said Brent Taylor, executive producer of GO WEST and president of Timewise Event Management. “Let’s embrace it and have those conversations around privacy and appropriate use. We are thrilled to engage with industry leader Zenus to showcase these technologies to our community of event professionals.”

Zenus CEO and Co-founder Panos Moutafis, PhD, will also present on the use of machine intelligence and facial recognition during the event.

Go West 2020 will be held January 19 and 20 in Edmonton, Alberta, Canada.

At the 2019 World Artificial Intelligence Conference (WAIC), Telpo received significant attention for its facial recognition products, including coverage of its Telpo Smart Terminal on leading Chinese news and opinion network CCTV, according to a company announcement.

Telpo facial recognition terminals are used by a community elder care service center in Shanghai for visitor registration, increasing efficiency with automated form filling and improving care for people with memory or other self-care challenges. They are also deployed to ease hotel and event check-in procedures, the company says.

Shenzhen has launched a facial recognition access control system to secure entry to its subway system, South China Morning Post reports.

Facial biometrics from Tencent grant free entry to registered people over the age of 60, and the service is expected to be extended gradually to other groups such as military veterans. The subway service has deployed 28 automatic gate machines and 60 self-service ticket processors to 18 stations on the city’s Line 11.

A ticketing system to test payment with facial recognition was also installed at Futian station in March, replacing cash or QR code payment. A similar system has already been in production at Jinan since April, with 500 commuters reported to be using facial recognition each day.

Similar systems are in place in about ten other cities, according to the report.

SCMP cites a January study from the Payment and Clearing Association of China, which showed 85 percent of respondents were willing to use facial recognition for payments, while the number worried about leaks of personal data fell to 73.8 percent from 84.8 percent a year earlier.

Article Topics

accuracy  |  algorithms  |  biometric testing  |  biometrics  |  facial recognition  |  NIST  |  retail biometrics  |  Telpo  |  Tencent  |  Zenus