10 Sep Feature: Facing up to biometrics
When it comes to biometrics for security there is little doubt that facial recognition is now at the very forefront of ‘frictionless’ ways to authenticate the identity of individuals. Timothy Compston casts his eye over the latest developments
We are seeing the latest generation of powerful facial recognition solutions being brought to market that offer higher speeds and a greater degree of accuracy than ever before, thanks to advances in computing power – AI (Artificial Intelligence) and, specifically, a deep learning architecture – plus the move from 2D to 3D to cope with a wider range of viewing angles and lighting conditions. Crucially, this facial recognition revolution is helping to unlock a growing footprint of applications, whether it be: to ramp-up security for mobile banking; to secure smartphones; to control access at key facilities or, when combined with video surveillance footage, to tackle crime in our smarter, safer, cities.
The speed of change here is breath-taking with a new report from The Carnegie Endowment for International Peace pointing out that at least 75 countries around the world are actively using AI tools like facial biometrics. In terms of specific figures on the dramatic upswing in the value of the global market for facial recognition, the research firm MarketsandMarkets is now forecasting a more than two-fold increase in value from US$3.2 billion in 2019 to US$7.0 billion by 2024 at a CAGR (Compound Annual Growth Rate) of 16.6 percent. MarketandMarkets points to the growing surveillance market, rising government and defence deployments, and technological advances across various industry sectors, as the key drivers for growth here.
Recent analysis released by Adroit Market Research is even more bullish, predicting that the worldwide facial recognition market will exceed US$12 billion by 2025. Where Adroit agrees with MarketsandMarkets is in the prediction that 3D facial recognition will have the largest slice of the market by the end of each forecast period. When touching on why 3D has the edge over 2D, MarketsandMarkets attributes this to the fact that this technique makes use of facial contours to identify and analyse various unique features in a human face. In addition, MarketsandMarkets notes the ease in detection of facial data from videos and 2D images and the way that 3D is least affected by illumination issues.
Middle East deployments
Focusing on the Middle East market specifically, facial recognition deployments are now wide-ranging with key territories and sectors ahead of the adoption curve. On the banking front, for example, last September, Aion Digital, one of the fastest growing FinTech solution providers in the GCC, announced a partnership with Daon – a global leader in biometric identity technology – to introduce the IdentityX two-in-one platform for digital customer on-boarding. Rewinding to 2016 and Gulf Bank Kuwait asked Daon to provide biometric authentication within Gulf Bank Kuwait’s mobile banking application via Daon’s Identity Platform. By working with Daon the bank’s customers were then able to login with biometrics using their fingerprint touch ID and ‘Blinking to Bank’ facial recognition, from anywhere in the world, to perform a wide array of banking transactions efficiently and securely.
Outside of banking, take-up has accelerated elsewhere, in another step aimed at making ride-hailing a completely safe and secure experience for its users, in April 2017 we witnessed Dubai-based Careem, the region’s leading ride-hailing app, announce the integration of facial recognition technology into its technological framework. Powered by Digital Barriers, the back-end biometric identification system would enable Careem to confirm its ‘Captains’’ identity in real-time, eliminating all of the risks associated with fraudulent car ownership and possession. Dubai International Airport (DXB) too has been leading the way for facial recognition with Princeton Identity’s multi-modal biometric technology is being deployed in the Emirates Airlines Terminals. This is in the form of its Access500e identity management kiosk module, a fast face and iris biometric capture device.
Eye on crime
It is also interesting to reflect on the announcement by Dubai Police that, under a new Artificial Intelligence (AI) network, thousands of video surveillance cameras from Dubai government agencies are now going to provide a live feed to one central command centre. The Oyoon (eyes) initiative applies AI and facial recognition technologies to help spot crimes and incidents by analysing live video, with no human intervention, and enables police to track criminals across the city by uploading their image into a database. Implemented over the past two years, the initiative is part of the Dubai 2021 Vision of a smart city and preparations for Expo 2020.
The value of video surveillance cameras fitted with facial recognition technology and also vehicle licence plate readers in a city environment was underlined at the end of 2018 when it was reported that these type of cameras had helped Al Muraqqabat police station in Dubai to arrest 100 wanted people and 441 suspects that year. Other police forces are investigating the utility of facial recognition with London’s Metropolitan Police Service confirming last month that it is to begin the operational use of Live Facial Recognition (LFR) technology from NEC. This it points out will be ‘intelligence led’ and deployed to specific locations in London to tackle serious crime like those involving guns and knives.
There has been some push back in the UK, the US, and Europe by privacy campaigners with regards to facial recognition. Interestingly the first report from the AI and Policing Technology Ethics Board set-up by Axon – the body-worn camera vendor – said that: “Face recognition technology is not currently reliable enough to ethically justify its use on body-worn cameras. At the least, face recognition technology should not be deployed until the technology performs with far greater accuracy and performs equally well across races, ethnicities, genders, and other identity groups.” Beyond this some US cities like San Francisco have said they will not allow facial recognition on their streets and the European Commission is said to be considering a ban on the use of facial recognition in public areas for up to five years.
Returning to Dubai, on a more positive note, facial recognition technology was certainly much in evidence at this year’s Intersec exhibition. There was, for example, the announcement of a strategic partnership between VIVOTEK – the IP surveillance solution provider- and CyberLink Corp – a pioneer of AI and facial recognition technologies. This will see the integration of CyberLink’s FaceMe AI facial recognition engine into VIVOTEK’s IP surveillance solutions. Commenting on the move, Owen Chen, chairman of VIVOTEK, said: “We are honoured to partner with CyberLink, also a Taiwanese company and to adopt CyberLink’s FaceMe, which was ranked one of the most accurate AI facial recognition engines in the [U.S.] NIST [National Institute of Standards and Technology] Face Recognition Vendor Test (VISA and WILD tests).” He added that: “Through this strengthened alliance, facial recognition intelligence will be integrated with VIVOTEK’s network cameras and back-end video management software, enabling security operators to receive accurate facial recognition alerts based on both blacklists and whitelists.”
For its part, Hikvision used Intersec 2020 to showcase multi-intelligence cameras that are designed to run several deep-learning algorithms in parallel for a host of complex scenarios. These cameras are said to be able to visualise and analyse structured data of various targets such as faces, bodies and vehicles simultaneously. Deep learning and facial recognition were also in the picture for Dahua in the shape of WizSense, a series of AI products and complete end-to-end solutions based on deep learning. In a perimeter protection scenario, Dahua notes that, in addition to recognising whether a target is a person or a vehicle, WizSense is equipped with face detection and face recognition features to aid the further identification of targets by comparing face images with those recorded in a customised database.
Also, at Intersec, Suprema Inc. – a leading global provider of biometrics and security solutions – unveiled its latest facial recognition solution. Suprema reckons that the new iteration of FaceStation – an access control terminal – which will be released later this year, represents a significant improvement over existing face recognition products. The company says that it is anticipating high growth in the face recognition market as it receives enquiries from customers prior to the launch of the latest FaceStation in the second half of 2020.
Away from Intersec, on the consumer front, facial recognition is now very much an integral element of smartphone security. Apple, for example, went for facial recognition in a big way with the launch of the iPhone X which featured Face ID to enable an individual’s face to be employed as a password to unlock, authenticate, and pay. Moving ahead a report by Juniper Research suggests that by 2024 facial recognition will feature on 800 million mobile devices, including 90 percent of smartphones.
Returning to the debate over the relative value of 2D or 3D facial recognition, advocates of 3D say that unlocking a third dimension in facial recognition means that these solutions should be better able to cope with a range of viewing angles and even changes to people’s appearance, from facial hair to sunglasses. A case in point is the approach taken by Artec ID – the 3D biometric division of Artec Group – and its Broadway 3D system which provides a very accurate mesh – structured light pattern – to make high speed recognition possible, against templates stored in a database, in less than a second. Developed for recognition on the move, according to Artec, the Broadway 3D system’s ability to read information about face shape has been optimised for high throughput facilities like office buildings, factories, and transport hubs where being non-contact makes it a ‘very friendly’ process.
An important development for facial recognition in recent times has been the advent of deep learning. With modern deep learning techniques, facial recognition engines such as SmartVis from Digital Barriers – that develops edge-intelligent solutions for the surveillance and security market – can identify users in most lighting conditions, even when they are not looking directly at the camera. As well as this, Digital Barriers – which was on the Nukleas Integrated Security Solutions stand at Intersec 2020 – says that off-the-shelf cameras can be used, dramatically lowering the cost. The ability of solutions like those offered by Digital Barriers to leverage existing video surveillance cameras – and not have to rely on bespoke hardware or special calibration – opens up the potential, says the company, for facial recognition to be deployed on a much greater scale than before around areas like critical infrastructure and crowded places.
Another example of the additional intelligence and computer processing power now being applied to this form of biometrics for border control, access control, and time and attendance, comes in the shape of facial recognition innovator Aurora. The vendor has a proven track record when it comes to state-of-the-art biometrics having cut its teeth in applications like London Heathrow’s Terminals 2 and 5 for airport passenger management. Specialising in identity verification for over a decade now, Aurora has moved from visible light systems through to pioneering the use of near infrared sensor technology to overcome the problems of varying ambient light conditions on a scene. The application of deep learning based Artificial Intelligence (AI) has proved to be a real ‘game changer’ for Aurora’s solutions. The key factor about deep learning in this field, according to Aurora is not just the scale of the performance enhancement but the speed at which it can be achieved. According to the vendor, the Neural Network, or ‘brain’, which it has developed to recognise faces has cut down the development time required to make significant performance gains from months to weeks.