Facial recognition: a hype, security risk or promising new tech?

Talk to the face

In recent years, improvements in AI and machine learning have prompted the rise of a relatively new type of tech that most of us call facial recognition. This new technology has the potential to replace older methods of people identification, like fingerprints, iris scans, voice recognition, handprints, passwords, and even physical ID cards – leading to significant process improvements and efficiency gains across sectors. The end goal of developing facial biometrics and using it to identify individuals is increasing security and minimizing the number of fraudulent transactions and cybercrime, which have become commonplace in the digital era.

As facial tech improves and becomes more accurate, face detection is expected to disrupt various industries, including private and public surveillance and security, forensics and criminology, banking and finance, and public administration, to name a few.

Inside facial recognition technology

How exactly does this innovative technology work? Facial biometrics are enabled by a combination of hardware and software, as most other tech. A 2D or 3D scanner captures a person’s face, along with all its features, shapes and peculiarities, producing a virtual model that’s very detailed. This model, with all its unique dimensions, is then transformed into empirical data by an algorithm. This is necessary for the model to be easily juxtaposed against billions of others ‘faces’ in a database, producing a reliable match in a matter of seconds.

The factors that go into constructing these sophisticated models are highly personal facial dimensions like the spacing between the eyes, the circumference of the nose, the distance between the lips and the chin, the contour of the lips, etc. Advanced facial detection tech companies claim they are already capable of verifying the identities of millions of people in crowded and unstable environments, using this model.

Apple iPhone X’s most publicized new feature is based on similar technology and it’s already accessible en masse for consumers to use. The controversial Face ID is based on the TrueDepth image system built into the device, which is comprised of an infrared projector and an infrared camera, capable of capturing all the data needed for constructing a personal facial model of the phone’s owner.

Facial biometrics implications

As facial tech become more reliable, its practical and business applications become more feasible – easier and cheaper. On the positive side, this means solving crimes is going to get easier, and finding the responsible parties – quicker, regardless of their geographical location. Satellite facial detection systems have already been utilized by U.S. intelligence agencies to spot and capture high-profile, wanted individuals. On the flip side, tech-savvy organizations can capture visual data publicly available on the web and social networks and exploit these photographs for commercial or more sinister purposes.

Face detection tech inaccuracies

Just like one’s voice or fingerprints, facial features can’t be altered and are thus a reliable personal identifier.  Despite being labelled one of the most promising new technologies around, facial recognition software and hardware are still in their infancy. Their accuracy has a long way to go before we can rely on them for important decisions, such as those concerning someone’s culpability.

In a striking example, on top of the extensive CCTV systems that are commonplace in London, the Metropolitan Police has been trying to implement facial recognition in people identification operations, though attempts have largely failed so far, at a rate of 90% or higher. The same ‘’success rate’’ has also been recorded at police departments in South Wales. The technology used by the British police has been dubbed the “fastest and most accurate face recognition technology.” That’s an alarming statement, considering how inaccurate it has been for these institutions.

Goodbye, privacy

Dystopian sci-fi films where giant electronic billboards display personalized advertising based on the profile of the person walking by them are no longer far-fetched visions of the future. Automated facial recognition devices, including supercharged public surveillance cameras, can identify and record the movement of people in virtually all public environments. If used carelessly, passersby could easily turn into walking ID cards, which carries its own security, privacy, ethics and commercialization risks.

In the next decade governments and public institutions will be hard pressed to introduce and instate regulations to govern the collection, storage and usage of facial and other personal biometric data, along with managing the awareness and consent of the public in doing so. Technology manufacturers, in turn, will need to ensure data isn’t made available to third parties or stored unencrypted. Apple’s Face ID, for example, ensures that identification takes place on the user’s phone and facial biometric data is never stored on the cloud where it could be more easily compromised.

Just like the European General Data Protection Regulation (GDPR) and E-privacy laws are aimed at securing online user behavior data, financial data and other personal information, facial data needs to be safeguarded in a similar fashion. The consequences of not managing these processes effectively can prove costly.

Copywriter: Ina Danova