Payment Verifications Driving Facial Recognition

Article By : George Leopold

Software-based approaches to facial recognition are making headway, but hardware frameworks are seen boosting authentication and trust.

Let’s face it, facial recognition is on the rise as a biometric identification technique for verifying mobile payments, driven by applications like Apple’s FaceID implementation.

But those who track the payment authentication sector say facial recognition will continue to lag current biometric techniques like fingerprint sensors until more robust hardware implementations hit the market.

The inflection point appears to be 2025, according to Jupiter Research, which forecasts software-based facial recognition technology will by then exceed 1.4 billion users. If accurate, that prediction represents a 120 percent increase over five years.

Overall, biometrics are expected to authenticate more than $3 trillion in transactions by mid-decade, up from about $404 billion last year. The drivers include Apple Pay and Samsung Pay apps increasingly used for remote and in-store purchases.

Apple’s FaceID is fueling most of the growth in facial recognition technology. Driven by the pandemic, the company’s latest software release, iOS 14.5, includes a FaceID upgrade that allows users wearing masks to unlock their phones.

facial recognition

Those upgrades along with AI-based verification checks are seen as beefing up current software implementations that still lack authentication trust as spoofing exploits such as deepfakes and “synthetic identities” increase.

As a biometric identification technique, software-based facial recognition applies deep learning algorithms to analyze and store facial features. Those data are then compared to image databases to verify identity.

The technique is controversial as a street-level surveillance method, but growing steadily as a means of verifying identity when, say, unlocking a device or authenticating online transactions.

While greater use of AI techniques may boost authentication trust, Jupiter Research said hardware-based approaches to facial recognition remain the best way to propel the market still dominated by fingerprint sensors.

“Hardware-based facial recognition is growing, but the ability to carry out facial recognition via software is limiting its adoption rate,” said researcher Susan Morrow. “As the need for a secure mobile authentication environment grows, smartphone vendors will need to increasingly turn to more robust hardware-based systems to keep pace with fraudsters’ evolving tactics.”

While AI- and hardware-based facial recognition systems are seen making a dent in the biometric authentication market, Jupiter Research predicts facial recognition technology will be just one piece of future payment authentication schemes. Along with dominant fingerprint sensors, other elements of its recommended “multi-method biometric strategy” include voice recognition and “behavioral indicators” such as keystroke patterns.

Verifying payments via voice recognition—a technique mostly used by banks—is forecast to top 700 million transactions by 2025. Like facial recognition, however, voice confirmation also lacks the reliability required to authenticate identity and, with it, provide the necessary level of trust as mobile fraud grows via social media platforms.

This article was originally published on EE Times.

George Leopold has written about science and technology from Washington, D.C., since 1986. Besides EE Times, Leopold’s work has appeared in The New York Times, New Scientist, and other publications. He resides in Reston, Va.

Leave a comment