What A Face

UNC Wilmington researchers build new facial recognition technology with a wide range of applications, including health and national security.

WILMINGTON — Faces are everywhere. In one day, you see the same faces of your peers, your co-workers and you might even meet a new face or two.

What about a place where you see a lot of faces at once? For example, the North Carolina State Fair. There are many faces there: young and old; happy, and maybe sad; and almost every age and expression in between. And all those faces tell stories about each individual.

“Your brain goes through an entire calculation when you look at a face and you try to identify that individual,” explains Dr. Karl Ricanek, professor at the University of North Carolina Wilmington and Director of the Face Aging Group Lab. “It says male or female, and roughly an age, and roughly a race, and then it says do I know anybody who looks like this age, this sex, this race, and then the brain begins searching its database for faces like that.”

It’s a basic description, but in effect that is how you pick out a face in a crowd. It’s called facial recognition. That’s what prompted Dr. Ricanek to wonder whether a computer could do the same thing, and perhaps even more.

“So my colleagues and I started out modeling the face and then we discovered there’s all this other information that I can extract from the face,” says Dr. Ricanek as he points to a computer monitor displaying many faces. “And when you pair it with identity, you improve facial recognition tremendously.”

It’s called facial analytics. 

Researchers at the Institute for Interdisciplinary Studies in Identity Science at the University of North Carolina Wilmington have studied the face for 12 years. They’ve discovered that unhealthy habits over time, as well as certain chronic conditions, leave their markers on the face. In short, understand the factors affecting health and soon you’ll find and understand the changes on the face that correlate. Researchers mapped 250 landmark points on the face. They’ve developed software so computers can create features maps and analyze the findings, comparing a test subject’s face to others in their category.

The effects of smoking are a good example.

Dr. Ricanek discovered that the face of a heavy smoker begins to show wrinkles in very specific areas. “And if you understand the biology of smoking it makes sense,” says Dr. Ricanek. “That’s because smoking literally dries out the dermis layer, and so what happens is that you start to prematurely wrinkle in certain areas.”

Health changes like weight loss can also be spotted based on facial characteristics. So the new focus is on whether the face can be used as a health care tool, and whether that tool can predict life expectancy as well.

To get an even better idea of how the facial analytics software works, I sat down in front of the camera to see how it would capture something very basic: my expressions. I centered my face in the face detection viewer box on the screen. In a matter of seconds, red marks appeared on my eyes, nose, cheeks and lips.

“There are a set of muscles on your face that keep moving when you have a different expression,” explains Amrutha Sethuram, a researcher in the Face Aging Group lab. “For example, if you have joy there is a specific set of muscles which get engaged. This software is designed to detect joy, sadness, surprise, neutral, or fear or disgust.”

The areas of my face that are highlighted in the face detection box are the landmarks that move according to the emotion. Once the landmarks are found, the face is registered and then the movement is measured. Because certain parts of the face register joy or disgust, the red points are used to detect certain action features. The degree of movement corresponds to the emotion.

So I then laughed, a lot. Then I changed expression, and looked sad. I also tried neutral, although I admit, I wasn't exactly sure how to look "neutral."

Sure enough, in the chart on the screen next to the window, the software registered joy, and sadness. When I attempted a lack of expression, the system read it as neutral.

Dr. Ricanek says the beauty of the face is that in addition to identity, the face holds a lot of information; he hopes to use that information for good.

“I see nothing but how we can help humanity with this technology,” adds Dr. Ricanek. “My goal is to use the face and other technology to create a health index that can be made available to anyone, and all that we would need is a smartphone and a selfie to take advantage of it. Upload the photo and after a few computer key strokes you could understand what your health index is and then start to make changes in your life. That would be a life changing technology!”

Related Resources: