- Thierry Spanjaard
Facial recognition on the path to recognition
Facial recognition has been around for a few decades already. The idea of automatically recognizing a person started at the same time as photography in the mid XIXth century. However, the technology was far from being ready, and research became more efficient on this topic only in the 1980s, at the same time as other biometric aspects (fingerprint, iris, etc.)
As often happens with new technologies, the first developments were targeting military applications, especially when the US were engaged in the Afghanistan and Iraq wars. The principles were already the same as now: a face is decomposed into a series of points and metrics that are in turn transformed into digital values. Recognizing a face consists in comparing these values with the ones stored in an existing database of faces.
The secure transactions industry has built an expertise in biometrics and is applying it in all market segments. Facial recognition is now used in payment apps both at enrolment stage and at payment stage. All major players in our industry, Gemalto, Idemia (the new name of the Oberthur Technologies – Morpho combination) and Giesecke & Devrient all have an offer in which they use facial recognition to provide financial services. High quality cameras now equip all smartphones, algorithms have been improved, and computing power is available both on mobile devices and in the cloud. Also, Pay-by-selfie has been demonstrated by all leading players and organizations such as HSBC and MasterCard have already introduced selfie ID verification on their mobile apps.
However, many end-users are still not ready to accept face recognition. Morning Consult, a US-based market research company, just published a research showing that while 34% of respondents are very favorable or somewhat favorable to “facial recognition software in personal devices,” the percentage of respondents who are somewhat unfavorable or very unfavorable reaches 39%.
A reason of the reluctance to adopt facial recognition could be that we unconsciously consider that recognizing our lovers, friends or business acquaintances is inherently human. Here too, research demonstrates the contrary. Already a couple years ago, Baidu, a major Chinese internet company, announced its facial recognition technology was performing up to 99.77% accurately, making it better than humans. Also, it has been demonstrated that Facebook is able to tell with 97% accuracy if two pictures represent the same person or not!
The most important step on the path to acceptance is probably the adoption of face recognition by Apple as a means to unlock its latest smartphone series the iPhone X, which is to be available in a couple months. We can already bet face recognition will get a new acceptance by the public at large when it is in everyone’s pocket.