Giant leaps forward in accuracy have led to more diverse applications, but how will privacy and data protection rights be upheld?
Recognising each other’s faces is part of what makes us human – there’s a whole section of our brain devoted to facial perception. But rapid development in the artificial intelligence (AI) behind facial recognition systems means it can now achieve 99.9% accuracy, as claimed by Unisys, the system used by the US Customs and Border Protection agency. And an improvement in accuracy means a boom in application, with the global market for facial recognition systems expected to grow from £2.5bn in 2019 to £5.4bn by 2024.
Facial recognition is biometric software that can identify or verify a person from a digital image by mapping out their features mathematically and saving the information. Using deep learning algorithms, it compares images to ensure the individual’s identity, similar to identifying technologies for fingerprints and retina scanning. A visual search engine tool, using a database of pre-identified people, can pick out key factors even in a crowded environment.
The technology is already being used for security, health and retail. Apple’s iPhone X, launched in 2017, popularised Face ID to unlock it, and in October 2019 Google Pay announced payment approval on an Android device by facial recognition, which claims to be more secure than passwords. Nonetheless, this year Google came out in support of the EU’s proposed ban of its use in public spaces (see box), saying face-related technologies must be developed responsibly.
Google highlights the difference between face detection (is it a face?), recognition (whose face is it?) and clustering (is this the same face in these other images?).
Facial recognition is used to combat crime and terrorism, with some systems able to find 100 people in a single image and match them against databases of tens of millions of faces. Police in China routinely wear smart glasses with integrated facial recognition software. In the EU its use is tightly controlled – the ‘man in the hat’ responsible for the 2016 Brussels terror attacks was identified by FBI software – while the German government is planning to use it at stations and airports.
As part of a UK trial, South Wales police used it at the 2017 UEFA Champions League Final – largely unsuccessful – and at the world’s largest Elvis festival in Porthcawl, in September 2019, where around 35,000 fans had their faces scanned in a bid to catch troublemakers and criminals.
The technology promises opportunities in customer behaviour analysis. Since 2017, KFC’s Chinese outlets and Chinese retail giant Alibaba have deployed a facial recognition payment solution that predicts customer orders based on previous choices, age and mood. In the travel sector, Beijing’s Daxing International Airport – opened in 2019 – uses the technology at its self-service boarding area.
And in healthcare, it is being applied for patient security and identification, as well as in diagnosis by detecting nuances in facial expressions associated with pain.
Supply chain applications
As accuracy improves, there are also clear opportunities in a number of areas. A joint venture between JDA and Panasonic has set out to prove that the use of real-time sensing technologies in factories and warehouses can improve efficiency and enhance the value chain of logistics and retail operations.
The EU is considering a three-to-five-year ban on the use of facial recognition technology in public places, while in September 2019, a UK court ruled government use of the tech does not violate privacy and human rights.
And after real-life scenarios such as facial recognition being used against protestors in Hong Kong, many believe there is a need to control its use to protect civil liberties.