What is facial recognition?

2022 is upon us. It’s that time of year where we run a series of blog posts on the questions most frequently asked by you, people of the internet. Something to keep you entertained during the grey, January period.

Let’s kick this series off with a Chapter One question:

It’s quite possibly the most important question that we receive as a company. 

Facial recognition is still a misunderstood and poorly represented technology; it doesn’t receive the recognition that it deserves.

What is Facial Recognition?

Facial recognition is a technology that uses artificial intelligence (AI). 

An algorithm is trained to complete a task. This could be anything, from arranging a dataset, to predicting the weather in Wisconsin. 

In our case, this task is recognising faces from a dataset.

The better you train your algorithm, the more likely it is to complete the task effectively. 

For our product Facentry, this is as simple as: sees face, recognises face, opens a door.

There’s a lot of complicated technical things that happen behind the scenes to enable this action to take place – but in short, the software itself focusses on authenticating that it does, in fact, see a face in its view, and that this face is allowed to access this door. 

Facial recognition uses biometric data to analyse faces. For TouchByte, this means that when a user uploads a selfie, it is verified as a suitable ‘biometric’ selfie. 

This image is then turned irreversibly into code; it stops being John, the 49-year-old who’s approximately 6”3, and is instead converted to a series of numbers. Numbers that measure the space between your eyes, the contour of your nose; all the little things about your face that make you, well, you. 

There are different types of facial recognition, which are used for very different purposes.

LFR vs Facial Authentication

Facial recognition means lots of different things to lots of different people. It’s a widely applicable technology which is useful in several sectors – all to achieve very different results.

You may have seen it implemented in environments such as border security or passport control, to identify and safeguard. Or, perhaps, accessing super-duper top-secret space buildings with rockets inside, destined for the moon.

Or, perhaps, to simply unlock your iPhone or to pay for your morning cappuccino. 

Because it has such a wide scope and variety of different applications, the lines can easily be blurred between facial authentication, and Live Facial Recognition (LFR).

In short, they are very different technologies with very different applications in the real world. 

Authentication with face, for actions such as iPhone unlocking or accessing a space with your face, works by recognising a face it knows (a face enrolled to a system). 

This face has given its consent to be enrolled onto a system and is often put in place to create an added layer of convenience for its user. Can you remember a time when you didn’t unlock your phone with your face? Or use Face ID to access your online banking? I certainly can’t. 

LFR, on the other hand, has a different purpose and therefore runs on a slightly different ethical code. 

Instead of being used on a personal basis – something a user can opt out of if they choose to - LFR acts as a more intelligent CCTV. 

Live Facial Recognition is more often used as a surveillance tool, not to streamline an everyday process, like unlocking a phone or a door. It’s often used in environments to monitor large groups of people – think airports or large public spaces. 

This form of LFR frequently hits the headlines with questions surrounding ethics. Often, because the technology hasn’t been implemented in its optimal conditions, meaning it isn’t as accurate or efficient. 

The long and short of it? We don’t use LFR. Our facial authentication products require an opt-in from the user and are designed to create ease and convenience. Not to monitor. Not not control.

We simply believe in convenience, safety and security. 

What are the key takeaways from this article?

• Facial recognition and authentication are forms of AI (artificial intelligence).

• Algorithms need to be trained to complete their tasks. How well you train your algorithm will dictate how successful it is. Equally, exposing a facial recognition algorithm to a variety of faces (genders, ethnicities, and ages), will create a more ‘rounded’ algorithm. What it sees, it will learn to identify. Time and time again. 

• There are lots of uses for facial recognition, and they’re split into two groups: Live Facial Recognition (LFR) and facial authentication. 

• We believe in convenience, safety and security. Our products are created to authenticate. 

• Spending your Christmas watching Love Actually is probably more interesting than scouring the internet for information on facial recognition. But here you are.