MSU researchers investigate how aging can affect facial recognition systems
Michigan State University researchers are trying to determine to what extent aging can affect the performance of automatic facial recognition systems and how it will ultimately impact the technology’s ability to successfully identify criminals or ascertain when identification documents need to be renewed, according to a report by MSU Today.
For the study, which was conducted in partnership with National Institute of Standards and Technology, biometrics expert Anil Jain and doctoral student Lacey Best-Rowden discovered that 99 percent of the face images they tested can still be recognized up to six years later.
“We wanted to determine if state-of-the-art facial recognition systems could recognize the same face imaged multiple years apart, such as at age 20 and again at age 30,” said Jain, University Distinguished Professor of computer science and engineering. “This is the first study of automatic facial recognition using a statistical model and large longitudinal face database.”
However, the test results also revealed that the natural changes that happen to a face over time as an individual ages can lower the recognition accuracy rate if the images of the individual were captured more than six years apart.
This drop in face recognition accuracy is dependent on a person to person basis as some individuals tend to age more rapidly than others based on several factors, including lifestyle, health conditions, environment and genetics.
“This research shows the importance of capturing new images every four to five years to reduce the number of false positives or chance of not finding a candidate in a facial recognition search due to length of time between captures,” said Pete Langenfeld, manager at the Michigan State Police’s biometrics and identification division. “Criminal acquisition is dependent on the number of times a person is arrested, as the majority are not required to update their image. However, civil applications that require updated facial images should look at reducing the time between captures if it is greater than every four years.”
The researchers studied 23,600 images taken from two police mugshot databases of repeat criminal offenders. Each offender had at least four images of them which were captured over a minimum of a five-year period.
The decision to use mugshot databases was based on the repositories being the largest source of facial aging photos available with well-controlled standards to ensure the uniformity of the photos.
The study marks the largest facial-aging databases analyzed to date in terms of the number of subjects, images per subject and elapsed times.
“This comprehensive study by Jain and Best-Rowden provides for the first time an unprecedented body of knowledge regarding the limits of automated face recognition,” said Brendan Klare, CEO of facial recognition developer Rank One Computing.
Jain and Best-Rowden’s paper will be published in the upcoming IEEE Transactions on Pattern Analysis & Machine Intelligence journal.