DHS Biometric Technology Rally early findings suggest improved recognition of mask wearers
The first phase of the 2020 Biometric Technology Rally took place at the Maryland Test Facility (MdTF) in September and October.
The event focuses on assessing the precision of biometrics systems in correctly assessing the identity of people wearing masks.
Now, the U.S. Department of Homeland Security is releasing the first details regarding the 2020 Biometric Technology Rally, including the number of platforms and algorithms tested.
The first part of the event focused specifically on evaluating the acquisition systems based on their ability to correctly take images of volunteers with and without masks.
The test, which analyzed six face and/or iris acquisition systems and 13 matching algorithms was performed on 582 volunteers from 60 countries, and also measured volunteer processing time, and overall volunteer satisfaction.
The second part of the event will take place in January 2021 and will test the matching algorithms’ ability to accurately identify each volunteer wearing (or not wearing) a mask using images from each acquisition system.
The accuracy of face recognition algorithms capable of identifying mask wearers has improved substantially during the pandemic, with a 25 percent improvement since March shown by the latest NIST report.
However, there are calls from industry experts to evaluate these new applications cautiously.
Biometrics Institute urges caution in deploying face recognition solutions for mask wearers
According to the Biometrics Institute, the increasingly widespread use of masks should fundamentally change how identity is managed and the way facial recognition algorithms are developed and applied.
The experts quote the most recent NIST report, saying that despite the improvements in identifying mask wearers, face algorithms are still consistently less accurate than in the identification of people not wearing a mask.
‘We welcome this research from NIST which enables our members to mitigate these issues, grow public trust, and continue to keep people safe,” said Biometrics Institute CEO Isabelle Moeller.
“However, a theme common to all NIST’s tests on face recognition is that each algorithm performs differently”, she added.
To mitigate these issues, Moeller thinks that companies should fully understand the limitations of their algorithm and thoroughly test their performance both in-house and through an independent laboratory.
“As new systems are developed, they must be independently tested before being presented to the public,” Moeller concluded.
The Biometrics Institute is also calling for the biometrics community to ensure that new good practices are promptly implemented and existing guidelines thoroughly followed.
These could include, for example, the creation of areas where people can temporarily remove their masks to be identified by biometric systems.
The Biometric Institute also released its Good Practice Framework earlier this year.