What Is Facial Recognition?

As technology grows at a rapid rate, many of us feel we are getting ‘left behind’. This particularly applies to our knowledge of online security. First of all, there were passwords, then PIN numbers and now biometrics, of which facial recognition is just one factor.

So, what is ‘biometrics’?

Biometrics is simply a technical term for measurements and calculations of the human body. It is data that proves you are who you are. Under the umbrella of biometrics not only falls facial recognition but fingerprint ID and iris recognition. These are all seen to be unique traits in humans.

How does it work?

Facial recognition works by mapping out specific markers on your face. This will include, for example, the distance between your nose and lips or width of your eyes. These are called nodal points. Just like our fingerprints, our faces are unique and not easily imitated.

Where is it used?

There are two ways of mapping faces. The more basic forms are used in situations such as unlocking your mobile phone or app filters. This is the first way.

Our mobile phones only need to recognise one face- yours, the consumer. In addition to this, we opt in to have our faces ‘scanned’. Each time we unlock our smartphone, the facial recognition software ‘looks’ for a face via the camera and compares it to the facial mapping that you have stored. If it matches, congratulations- your identity has been approved!

When using apps, the process may be even more straightforward. Theses apps simply look for a face- eyes, nose and mouth and aren’t specific to one individual face. Instagram and Snapchat are great examples of such apps. The filters that they provide will suit any facial features. As such, this is technically just software that finds any face rather than individually mapping them.

These methods use unchanging images, unlike individuals faces searched in public places and crowds.

This leads to the second type of facial recognition.

When used in masses, such as concerts, sports matches or demonstrations, etc., this is an automated form. Human faces can be captured within their surroundings and separated from other objects such as cars, lamp posts, or buildings. The images are processed using AI (artificial intelligence). The picture is then compared to hundreds of thousands of others stored in a database. Scanning huge crowds takes minimal effort using this method.

Any database may be used to compare images with. This could include social media profile pictures or from identity cards in some countries. There are a couple of requirements- photos have to be clear and pre-identified, to be compared successfully. With all of those components running, identifying a stranger is straightforward.

Deep learning algorithms are used in this meticulous process. This is not unlike Apple ID but clearly on a much grander scale.

What are the pros?
  • Facial mapping is a non-invasive and no-contact process, and images can be taken from a distance.
  • Police say when it is used in public spaces, it can help with help security and locate dangerous people/suspects.
  • It can be used to track time and attendance for employees, students and the like.
  • It can be a cheaper use of technology due to using less processing.
  • Individuals have limited capacity to thwart facial recognition unless their face is actively covered or partially. In contrast, this can also be a downfall.
  • Facial recognition in airports can be used to verify your identity, to unlock your phone or to even enter your banking apps.
Are there any downfalls?
  • The technology won’t work if conditions aren’t ideal or a face is at least partially covered.
  • Fingerprint and iris ID are generally more accurate.
  • Privacy campaigners feel it is a significant threat to our ‘civil liberties’.
  • People are not well informed, if at all, about how images are stored and used.
  • Many people have no idea that their image may have been used.
  • Clearview AI in the United States has been under fire for scraping thousands of images from social media profiles without consent.
What does the future hold?

It may be used in advertising- a quick scan of your gender and approximate age means advertisements can be tailored towards you. There is a rumour that it could even be used as a form of payment. Once your face has been recognised, payment will be taken from the account you are linked to.

Yet it has been widely agreed that people need more information about how this works and to realise any drawbacks about how it may develop in the future.

In May 2019, San Francisco, California, became the first US city to ban the use of facial recognition, citing it as ‘psychologically unhealthy’. Recent studies have shown that the technology can be extremely flawed. This appears more prevalent with women, transgender people and racial minorities.

In contrast, facial recognition in China is rife. It has already been rolled out in cities to spot and, subsequently, fine jaywalkers. It verifies students at school entrances and is alleged to monitor expressions to assess if one is paying attention. Any individual buying a mobile phone contract has to have facial recognition performed.

China has also come under criticism for the way it has tracked and controlled Uighur Muslims, being condemned as shameful. (In the name of anti-terrorism and national security.) This has raised serious issues about personal security and consent.

Trials in London have not fared as well as hoped. The Met Police Department consider it 70% effective in recognising faces, yet a separate study clearly differs in opinion. Professor Pete Fussey conducted independent research and found it was only 19% verifiably accurate. Quite a difference between accuracy rates!

Furthermore, human rights advocates have described the measure as another step towards a “dystopian surveillance state”.

One thing is clear, without firm regulations and laws, any form of technology is open to abuse. We all deserve to know where any parts of our information, physical or personal, are stored and for which purpose it is being used.