Smilepass is an exciting new UK-based biometrics company. Last month we reported how its facial recognition technology is being used by Kwara State in Nigeria to improve the deliver of healthcare services to citizens. We wanted to find out more, so we asked CEO, Grant Crow to share his thoughts with readers of the ID Bulletin…
In the past onboarding people has been seen as an issue to mass adoption of biometrics. Has this changed?
When biometrics depended upon hardware, this was a significant barrier. Mobile biometrics removes this barrier. Therefore the primary remaining barriers are around trust and compliance with legislation (which are linked issues). Trust is an issue with older consumers but less so with millennials. In a recent IBM report, The Future of Identity, it stated 75% of Millennials are now comfortable using biometrics and 87% of all adults saying they’ll be comfortable with these technologies in the future.
Biometric sensors (typically fingerprint) on consumer grade mobile devices are more about convenience that security. Are they now robust enough to make using them for security purposes viable?
Companies need to ensure that they use biometric solutions that offer the optimal balance of security versus convenience and cost. With facial matching as an example, accuracy can vary widely. There are definitely solutions available that are sufficiently robust to provide excellent security. Using more than one mode is also recommended for some use cases as this will invariably improve security levels without detracting from convenience. All this is if the provided the solution is well designed, of course.
You recently announced an exciting project in Nigeria where fingerprint technology wouldn’t work for some people in rural areas where they prints had worn. How does facial recognition adapt to the changes in a person’s appearance?
Most good facial recognition algorithms are capable of matching an individual without a beard or glasses to their picture with either or both. It is easy to resample an individual’s face over time and to store the most recent facial profile. In that way, changes are seen gradually and become less of a challenge. Of course this is also greatly dependent on the frequency with which people use the system. If they use it frequently then the changes are easier to adapt to as the system is resampling data every time they log in. If the changes are sudden and extreme it can present more of a challenge and the accuracy threshold may need to be reduced.
What are the key application areas for facial recognition technology today and in the future?
Today, many of the applications are focused on proof of identity for fraud reduction. Tomorrow will see more applications focused on attendance checking perhaps combined with geo-location. As the technology becomes more accepted and is proven to be fast and accurate, the use cases will evolve from the simple use cases we see now of identification, verification, and access management to numerous applications. As facial recognition technology matures it will become more of a commodity used within more complex applications such as venue tracking, retail usage analytics, and even tracking the flow of people through cities and transportation networks. All this will further reducing the amount of explicit authentication a person needs as they go about their lives, both digitally and in the real world.
What is the accuracy rate of SmilePass? Does it need to be part of a multi-factor authentication approach to be truly secure?
The accuracy rates vary between our different levels of security. At the lightest end of the scale, SmileCheck with its focus on attendance has accuracy rates of circa 85% whereas SmileSafe + has accuracy levels in the high 90% and therefore is sufficiently robust to stand alone. Accuracy also needs to be combined with non-spoofability to ensure that security is maximised. Accuracy thresholds are also affected by the nature of your application and how people use it. If used frequently our system can track how most people change, however if there is a sudden and extreme change to appearance we may need to register a user again. Generally speaking the more times a person uses our system the more accurate it will be, as we resample their biometric data every time.
For more information about Smilepass visit: https://smile-pass.com