Face morphing, where an attacker merges photos of two people into a single image, is a specialty of Mei Ngan, Computer Scientist, National Institute of Standards and Technology (NIST) and leader in morph detection.
“If that image gets onto an identity credential such as a passport, it can be used by both people because modern face recognition algorithms will authenticate and match images of both people to the morphed photo,” says Ngan.
That vulnerability led Ngan to create the NIST Face Recognition Vendor Test (FRVT) MORPH program in 2018, “to assess whether morphed images could be detected using software algorithms and how well they work.”
In addition to her expertise in morphing and face recognition evaluation, Ngan launched the Tattoo Recognition Technology Program in 2014. It was the first program at NIST to provide a measurement and testing foundation to support applications for image-based tattoo recognition. Her prior work at the MITRE Corporation set the stage for this program.
“At MITRE, I was part of a team that worked with the FBI to develop a tattoo dataset and a set of use cases that could be used to evaluate tattoo recognition technology,” Ngan says. “At the time, there were no common data sets or protocols to evaluate and develop operationally relevant tattoo recognition applications. So the goal was to create something that could help spur research and development in this area.”
There were no common data sets or protocols to evaluate and develop operationally relevant tattoo recognition applications … the goal was to create something that could help spur research and development
Ngan has served as the project lead for several major biometric evaluation programs at NIST, making her one of our 2020 Women in Biometrics award winners.
Ngan’s proudest accomplishment:
I’m a part of a team that conducts testing of face recognition on an international scale. One of the collaborations was with a U.S. government group that investigates child exploitation cases and works to recover child victims and to arrest perpetrators. We wanted to assess whether tools like face recognition would be able to help the analysts in their day to day workflow. So we stood up a testing capability at the U.S. government facility which housed the imagery, and then we ran a public evaluation where we invited the face recognition community to submit algorithms that might work on identifying faces of children under extremely challenging conditions. I’ve got children of my own, and I can say that I am quite proud to be able to make a small contribution to that mission.
The Women in Biometrics Awards, co-founded by the Security Industry Association (SIA) and SecureIDNews, will recognize five winners during the virtual 2020 SIA GovSummit June 1-4, 2020.