In this SEN Q&A we chat about considerations of face recognition technology with David Tullipan, ANZ sales director of biometric devices business line, at leading biometrics developer, IDEMIA.
JA: What are the most important aspects of successful face recognition applications, in your opinion? What should integrators and end users be on top of before they even start looking?
DT: Biometric terminals, with face recognition applications, are deployed based on a need for increased security to prevent intrusions, and once installed they will be ina operation for years. This means the 2 important aspects to consider are the quality and performance of the technology. The key selection criteria of face recognition technology include its biometric accuracy – where one should assess whether the vendor’s algorithms been evaluated by neutral, third parties like NIST or DHS. Secondly, integrators should look at its performance when utilised – such as speed, convenience, and resistance to spoofing attacks.
JA: Taking the diversity of facial recognition solutions – SME access control, enterprise access control, VMS AI, etc – in account, what general features do the best face recognition systems have?
DT: Simple connectivity to PACS and VMS is essential. There’s indeed a great variety of needs and corresponding solutions that can be customised and deployed in different environments. For example, terminals installed on speed gates help manage high throughput, terminals installed outdoors are built to work in strong sunlight, or complete obscurity for nightshift workers, as well as having a software solution plugged onto CCTV/VMS systems to manage live alerts and be used for access control as well. In all cases, accuracy, performance and reliability remain the 3 key features to look for in choosing face recognition systems.
JA: Would you argue face recognition – low touch, highly and increasingly accurate – has the potential to be ‘the’ authentication technology of choice in the future?
DT: Yes, indeed. Face recognition is a biometric modality that has grown in interest due to its ‘touchless’ feature. As many end users wear face masks today, biometric terminals and its algorithms are recalibrated for higher accuracy, to authenticate users despite having their faces covered partially.
From IDEMIA’s point of view, our VisionPass, combining 2D, 3D and infrared cameras with algorithms, have been recognised by recent independent evaluations, and is well known for its high accuracy and efficiency with masks worn on the face, even when worn with glasses. It also offers a mask detection feature that prompts an alert message on screen if the user is not wearing a mask. Besides face recognition, we also have MorphoWave contactless fingerprint scanning terminal that can scan one’s fingerprints with just a wave of a hand.
JA: What is the greatest challenge of installing and managing a face recognition solution?
DT: As a face recognition system requires deployment both physically and digitally, the installation process of terminals and its digital systems can be a challenge, if an end user is not partnering with an experienced vendor. For a smooth deployment, both from a timing and costing perspective, you need to select terminals that are already supported by the PACS system and speedgates/turnstiles that you are using or consider deploying.
IDEMIA’s terminals are supported by over 90 per cent of PACS vendors and the majority of gate vendors. The design of terminals is equally important – our engineers pay special attention to the installation, be it for gate or wall mounting and cabling.
JA: Is it fair to say that over time, a face recognition-powered access control system might pay for itself through circumventing the need to manage expensive card libraries?
DT: We see it differently – the business case of biometrics is really around your fight against hacking and intrusions. The access control credential should be determined by the security risk. Dual factor authentication (card + biometric) is more secure than single factor authentication. Ultimately, triple factor authentication (Card + Biometric + PIN) – also referred to as “something you have plus something you are plus something you know” – is most secure.
While users may not utilise access cards after biometric terminals are deployed, it is still a form of identification that is recognised by security managers and public areas. All of IDEMIA’s terminals work efficiently with cards, by putting the user’s biometric template into the card delivering 2-factor authentication with the device. In addition to that, the card can be saved virtually in your mobile phone. In the future, we can expect more 2-factor authentication being used for a diverse range of applications.
JA: When it comes to the underlying software algorithms that deliver face recognition, how can integrators and end users ensure they get the best of the best?
DT: Unlike other industries where IDEMIA operates, such as the payments space where cards and terminals have to get functional and security certifications from payment schemes, there are no such frameworks for biometric terminals. As a result, many integrators and end users look at independent evaluations carried out by entities like NIST or DHS. Integrators and end-users should assess whether the vendor had submitted its algorithms to such benchmarks, its ranking, and performance over time. Having concrete statistics and a detailed level of information about accuracy performances can help end-users better compare terminals and algorithms.
JA: There’s enormous flexibility in the application of face recognition analytics – not only for security but in terms of the ability to drive sub systems, report events and streamline investigations. In your opinion, what are the most exciting applications of the technology, and which might offer users the best return on investment?
DT: Flexibility is a key aspect of the video analytics solutions that we develop at IDEMIA. We research and develop new features to respond to the needs of companies looking to prepare for a safe and healthy return to the office. Augmented Vision is IDEMIA’s video analytics platform that helps protect buildings while ensuring people’s privacy with the highest level of data protection. The solution allows secure and frictionless access to authorized personnel by detecting and identifying multiple persons approaching an entrance, or point of access, placing a powerful layer on top of existing CCTV infrastructure.
The solution is also efficient with face masks, with the same features and performance as VisionPass. There are other features like physical distancing, contact tracing, overcrowding alerts which were previously not a customer priority more than a year ago. We believe there is a big gap to fill in post-event analysis. Augmented Vision is able to analyse enormous volumes of recorded videos and images, classifying and extracting all relevant information to help security officers increase productivity and efficiency. This is the biggest ROI, and it will help to resolve the challenges that companies with CCTV systems are facing at this point of time.
JA: Securing face recognition systems – what’s the key here – how can security integrators and security managers ensure private data stays private – or are privacy fears overblown?
DT: Companies that install face recognition solutions must adhere to the security recommendations to combat against cyber threats, such as proper authentication, access rights, encrypted communications, encrypted databases, network protection, and penetration tests. This is true for every system that manages personal data, not only face recognition systems.
Regarding data protection, individual biometric data like fingerprints or face images are not stored by IDEMIA. The biometric data is reduced and changed into biometric point characters, with their own template and biometric code. This, in turn, is secured before processing. Only companies which install these devices have highly secure access to this data, pursuant to current regulations.
We believe people need to know how their data is used, how long it is saved and why IDEMIA privacy-by-design solutions ensure that private data does not leak from government-authorized environments.
JA: How important is image quality when it comes to face recognition applied to CCTV systems?
DT: Very important! The image and camera quality required is dependent on the application and the distance of the person from the camera. For example, access control in a corridor where employees will be at maximum 2m away from the camera, is not the same as mask detection in a lobby where individuals are more than 10 meters away from the camera. Nonetheless, to achieve optimum performances in face recognition, we recommend having at least 40 pixels inter-eyes.
JA: How interested is the end user market when it comes to face recognition – is there hunger for the technology?
DT: We have seen that end-users have a massive appetite for touchless solutions, be it face recognition or contactless fingerprint scanning. Currently we see that end users remain vigilant on terminals’ accuracy levels and its ability to work efficiently with masks. Employee acceptance and compliance with data privacy regulations, which are 2 linked topics, are also a growing concern.
#sen.news
