A recently completed demonstration of biometrics technologies in high-throughput environments shows that the companies are making the systems easier to use and driving up user satisfaction, according to the Department of Homeland Security official in charge of the evaluation.
Compared with the first Biometric Technology Rally in 2018, this year showed a “significant increase in the number of really strong candidates,” Arun Vemery, director of the Biometrics and Identity Center at the DHS Science and Technology Directorate, told Defense Daily in a May 17 telephone interview. The systems that were demonstrated were “faster, more accurate and all rated really well by volunteers that interacted with the technologies too.”
Two key similarities between the 2018 and 2019 rallies that Vemery’s team evaluated were user satisfaction and how long it took for the volunteers in the study to interact with the particular biometric technology that was used at that time. He said the technology rally seems to be working as an effective way in engaging with industry.
Vendors made improvements and even “came in swinging for the fences and did quite well,” Vemery said. “So, there’s better options and more options to choose from.”
The throughput threshold for the technologies used was 10 seconds and the objective was less than five seconds, he said.
Results are preliminary and S&T is still doing additional analysis, Vemery said.
The 2018 and 2019 rallies were open to finger, face and iris data collection or systems that could collect a combination of the modalities. Most of the vendors in the recent rally brought facial image collection devices.
The rallies are conducted at S&T’s Maryland Test Facility in Upper Marlboro. There were 430 volunteers that went through the recent high-throughput scenarios with no training so the vendors were responsible for making sure their systems were as “intuitive as possible and accommodating,” Vemery said.
A new addition to the latest rally was the inclusion of 15 biometric matching algorithms. More than 13,000 images were collected in the latest rally and now S&T is running those against the algorithms before it completes the analysis portion of the study, Vemery said.
The analysis of the matching algorithms will look at how performance varies and how consistent it is when images are used from different cameras, and how well a particular algorithm will work with any camera system, he said.
Most of the matching algorithms provided were for facial images but several were for fingerprint and iris matching as well, he said.
In addition to providing the participating vendors with feedback, S&T plans to host webinars for its stakeholders across the public and private sectors to review the findings.