Avatar photo

Contributor

Author Bio ▼

Adam Bannister is a contributor to IFSEC Global, having been in the role of Editor from 2014 through to November 2019. Adam also had stints as a journalist at cybersecurity publication, The Daily Swig, and as Managing Editor at Dynamis Online Media Group.
September 23, 2016

Sign up to free email newsletters

Download

Whitepaper: Multi-residential access management – The move to digital

Biometrics and AI: how FaceSentinel evolves 13 times faster thanks to deep learning

Gary James showcases IDL Fastlane turnstiles with FaceSentinel biometric solution

Gary James showcases IDL’s Fastlane turnstiles with the FaceSentinel biometric solution at IFSEC 2016

Powered by ‘deep learning’, FaceSentinel is a world first in access control and biometrics, according to its architect, Aurora.

The system, which can be integrated with existing access control systems, uses artificial intelligence and infrared light to achieve unparalleled speed, accuracy and reliability, Aurora claims.

IFSEC Global met up with the company’s head of sales and marketing Gary James during IFSEC 2016, where FaceSentinel was showcased in conjunction with IDL turnstiles. You can read the conversation below.

IFSEC Global: So you launched FaceSentinel during IFSEC 2015. Twelve months later how is the product looking?

Gary James: We’ve used our deep learning expertise to advance the product to a new level.

So instead of just doing one-to-one verification, it will actually do identification. So this is moving away from tokens and cards and access control to just using your biometrics – your face – to perform a task. In this case, to open a gate.

We’ve partnered with IDL, so you can now actually see the products doing a job. And their technology within all these gates and door detectors works very well alongside ours, because it manages the whole process.

We’re opening the gate with biometrics, but the gates can protect against people tailgating in, which is a big problem.

So IDL have technology to detect or prevent that happening. So it works well along with our face sensor.

It’s taken nine months to produce the one-to-many. Developments of that kind we would expect to take about 10 man years without deep learning

IG:  How has the product been received at IFSEC?

GJ: There’s been a lot of interest in all these products working together. Because actually, the facial recognition systems we have working here are completely self-contained, because of the way IDL works. They have a different system architecture. So there’s actually no access control system involved in any of this; it’s completely standalone.

IG: What are the benefits of that?

GJ: You don’t need an access control system, so there are costs and benefits. But ultimately I’m guessing most users will have some kind of access control system for the rest of the building.

The point is, IDL have a different approach: it’s much more IP-based than a lot of these products. So we slot in quite nicely.

IG: How do you see the biometrics evolving? Is there still a perception that it’s unreliable or in some cases expensive?

GJ: I think that has to be changing, if you look at the way our biometrics are used and have been used for a long time.

We have a robust enough biometric to do time and attendance in the construction industry. That’s controlling the payroll for hundreds and thousands of people, across many different construction companies.

Probably the biggest indicator would be the adoption of the technology by Heathrow. We’re matching people to their boarding passes.

We believe it’s unique in that it’s the only biometric anywhere in the world where it’s mandatory for passengers to use it. At Heathrow it’s not optional to use our facial recognition. If you haven’t used it, you won’t fly.

It self-boards, so it’s proven to be more reliable than people. And I think once you have that level of robustness, frankly the access control task looks quite trivial compared to [other security elements at] an airport.

So we don’t see perceptions of reliability being a barrier anymore. Cost will always be I guess, because we use specialist sensors. But the high security market is often prepared to invest at that level.

IG: What does the artificial intelligence aspect mean in practice for users?

GJ: We use ‘deep learning‘, so it’s a piece of artificial intelligence powering it. The primary influence on the product is the speed with which we can develop engines and optimise them for different conditions.

So, for example, we launched products with the one-to-one matching last year, and it’s taken nine months to produce the one-to-many we have now. And developments of that kind of engine we would expect to take about 10 man years without using deep learning. So it’s a rapid effect on development. And it also increases accuracy and robustness generally.

IG: What kind of sectors do you target?

GJ: We already work with time and attendance and passenger management and many customers are in the corporate banking industry.

And we have systems controlling air-side access for staff at a major cargo company. They have staff that operate across many sites. One of the reasons they chose us is we can enrol people in London and they can work in Belfast without having to re-enrol. We’ve got that working across a European platform.

We see it used in a high security, corporate type environment. The one-to many is going to be a great way of increasing throughput in very busy offices.

Imagine the headquarters of a bank at 8:30am – it’s a busy place. People won’t be fumbling around with their cards anymore. That’s what we’re getting away from.

 

Subscribe to the IFSEC Insider weekly newsletters

Enjoy the latest fire and security news, updates and expert opinions sent straight to your inbox with IFSEC Insider's essential weekly newsletters. Subscribe today to make sure you're never left behind by the fast-evolving industry landscape.

Sign up now!

man reading a tablet, probably the IFSEC Global newsletter

Related Topics

Subscribe
Notify of
guest
0 Comments
Inline Feedbacks
View all comments