The National Institute of Standards and Technology (NIST) has released a study evaluating the performance of software that estimates a person's age based on facial characteristics in photographs. This age estimation and verification (AEV) software has the potential to serve as a gatekeeper for activities that are restricted by age, such as the purchase of alcohol or the access of mature online content. The study, "Face Analysis Technology Evaluation: Age Estimation and Verification (NIST IR 8525)," evaluates six algorithms that were voluntarily submitted by developers. Following its most recent assessment in 2014, this evaluation represents NIST's initial foray into AEV software analysis in a decade. It marks the beginning of an ongoing, new initiative by the agency to implement consistent and frequent testing of this technology.
In recognition of the increasing significance of face analysis software, NIST has structured its face recognition program into two distinct tracks: the first, the Face Recognition Technology Evaluation (FRTE), is dedicated to the identification of individuals, while the second, the Face Analysis Technology Evaluation (FATE), is dedicated to the measurement of facial aspects. The latest evaluation is part of the FATE track, which also includes assessments of photo spoof detection and image quality measurement. For this study, NIST expanded its photo collection to approximately 11.5 million images from four diverse U.S. government databases: the 2014 visa collection, FBI mug shots, webcam images from border crossings, and immigration application photos from individuals born in over 100 countries. This comprehensive approach aims to enhance the accuracy and reliability of AEV software, ensuring its effectiveness in real-world applications.
