Can identical twins fool facial recognition software?

Back in high school, the class ahead of me boasted a pair of identical twin boys. They dressed, talked and walked alike. They played the same sports. Their similarity fascinated me.

Identical twins are rare. According to the Centers for Disease Control and Prevention, about 3 or 4 births out of every 1,000 in the United States are identical or monozygotic.

The heroine in the book I’m currently writing is an identical twin who desperately needs to access her sister’s email. This led me to research whether twins can fool facial recognition software.

I learned that twins often trick basic 2D image-based systems because of their identical bone structure. These are the systems used on most cell phones and are famous for allowing twins — and even other siblings who bear a strong resemblance — to fool the facial recognition software.

On the other hand, the 3D imaging used in some programs is harder to confuse. It’s difficult to break into a Windows laptop when the owner has used Windows Hello: Microsoft’s biometric login for PCs. This software has multiple cameras that can usually distinguish between identical twins. In addition to 3D images, Windows Hello: also uses infrared imaging.

In the public, there are vast differences in the degree of accuracy of security scanners. Some often mistake individuals with similar facial traits. But high-end security and law enforcement systems can measure even minute details such as vein patterns and skin texture.

So how does facial recognition work?

Facial recognition converts a photo or video frame of a face into a mathematical code, then compares that code against stored codes to find a match. The process happens in a series of steps: detecting a face in an image, mapping its key features, translating those features into a numerical representation, and then checking that representation against a database. Most systems can perform the entire sequence in under a second.

Facial recognition software increasingly incorporates behavioral biometrics that analyze micro-expressions and other behavioral patterns unique to an individual. But this can still be a challenge when twins are involved, as they often unconsciously mimic each other’s expressions.

Hey, this sounds like an idea for another book!

The truth is, twins present a unique challenge to biometric identification, which often leads to misidentification. Some studies report error rates for twins to be several times higher than for the general population, but advancements in technology are constantly narrowing the margin of error.

Trying to thwart biometric software is serious, and the ethical implications are significant. Misidentification of twins can lead to identity theft or wrongful accusations in a criminal case. (Hey, this sounds like a great idea for a story!) Experts advise twins who want to protect their identity to take extra measures such as using a fingerprint or a strong PIN.

But back to my story

As for my former classmates, I had an opportunity to speak to their mom several years after we’d graduated. I mentioned that I’d attended school with her sons and had marveled at their similarity.

“Same here,” she said with a sigh. “I could never tell them apart either.”

Wow. So their own mom struggled to differentiate between them. That’s how alike they were.

I imagine the twins in my story are also so alike they’re hard to tell apart.

The question is will my heroine break into her sister’s computer by flashing her own face? Her sister’s in big trouble and the key to rescuing her may lie in that computer. Will the heroine succeed? I’m finishing the book as fast as I can, so you can find out!