Boosted PhotoVerify performance – New release!
Breakthrough for KYC Onboarding
Mutual exchange with our customers on the identity market and continuous development by the R&D team in Nuremberg have ultimately catered to our new PhotoVerify release which has boosted performance by factor 5 (!) of the beta version. This breakthrough allows a perfect balance between security and convenience for individuals and companies alike: The False Acceptance Rate (FAR) can be reduced to 0.1%, making it extremely hard for impostors to fake the system. The FAR in this case refers to fraudsters being able to perform a successful identification with PhotoVerify. Thus, a FAR of 0.1% means that, if you had the chance to try the identification 1000 times (which is far from a real-life scenario of course) you would only be able to surpass the technology once.
For customer satisfaction and retention, it goes without saying that at the same time is has to be ensured that the eligible users have a quick and easy experience. Therefore, we have also focused on delivering a very low False Rejection Rate (FRR), meaning, if you are the one on the ID, you will virtually always be identified as such. An illustrating example: With the new PhotoVerify release, even a 10-year-old ID document can be used to verify individuals by a selfie taken on their webcam. The robustness of the recognition can deal with these real-life situations so that neither a professional camera is needed for taking the selfie, nor is the age of the ID document or variance in appearance significantly reducing PhotoVerify’s efficiency.
BioID combines highly accurate face recognition with its stand-alone liveness detection, so the user’s presence is checked reliably to diminish any spoofing attempts before comparing selfie and ID photo. No matter if a fraudster tries to use a high-resolution image, a video or an animated 3D avatar, they will have no chance to spoof the identification. Most importantly the technology is not dependent on special hardware like 3D cameras to be scalable on the mass market.
Why online onboarding is an issue
The relevance of lean processes for registering new customers is growing as is their complexity. Onboarding traditionally meant a face-to-face conversation, a paper form to be filled out and a manual ID check to verify who the new customer was. It used to be best for customer satisfaction if you could offer an individualized helpdesk offline with a nice store to go to and employees knowing every customer’s name. Meanwhile, one of the strongest reasons to discard a business relationship or onboarding process is a sole offline experience, that can neither be offered 24×7 nor right at the customers’ desired touchpoints.
Often, what keeps businesses from predominantly moving online with their onboarding processes is the risk related to identity fraud and theft as well as various compliance issues, e.g. if mistakenly onboarding the same person multiple times. For instance, financial service providers have to make sure their (registration) processes comply with Anti-Money-Laundering (AML) and Know Your Customer (KYC) regulation, Payment Services Directive II (PSD2) as well as the General Data Protection Regulation (GDPR). These rules and standards create an area of conflict, defining which of the customer’s data is required and which is not to gather, how it is to be collected, governed and so on. Most importantly, of course, this system is set to secure individuals and companies alike from fraud and prevent personal data from misuse. Thus, a bank cannot simply trust the person registering online has a valid ID and is eligible to open a bank account without a proper identity proofing KYC process, which then become somewhat cumbersome, though.
So, while a high percentage of business processes could be digitized and established as an online offer in addition the offline one, the case of onboarding is still an unsolved issue. The interim solution of various video-ident processes tries to induce a face-to-face reliability with a video-chat and human agents manually checking every registering customer. While surely being a step in the right direction, this semi-manual process has not had the desired effect of solving the onboarding issue. It is cumbersome to use, dependent on a very strong and stable internet connection and anything but frictionless. In addition, the high potential of cost reduction for service providers through automation largely gets fed up through call-centers, fallback offline processes and other related issues. Still, until recently it was the only solution for online onboarding with high assurance levels.
Biometric Experience + AI = PhotoVerify
When analyzing the steps to be done for an identification, it turned out that with the help of advanced technologies like biometrics and artificial intelligence (AI), a more enduring solution than the video-ident process could be developed. The most important steps for an identification are the following:
(1) Genuineness check of ID document and (2) Validation of ownership of the ID document.
With (1) already having a vivid market of validity checking companies and methods, including Interpol database check and OCR readers, it soon was clear that step number (2) was the one to be addressed by innovations. As a biometrics company, BioID was able to assess the most important aspects to be considered for a promising solution:
(2a) Biometric verification of ownership through face comparison and (2b) User presence check to go sure, no fake was presented.
When designing an online identification tool, the performance of such a technology is dependent on various factors. Most importantly the biometric recognition as well as liveness detection have to be fake proof and highly accurate. At the same time a system to be used in unsupervised online onboarding has to be robust to external factors like lighting, age of ID document or visual changes of the person. Also, one of the main critics for traditional video-ident processes is the required bandwidth which is responsible for many failed registrations. Therefore, an enduring solution cannot be dependent on video-streams which all require strong internet connection. Additionally, it has to offer an excellent user experience, high success rates while at the same time preventing fraudsters from challenging the system.
With 20 years of experience in face recognition and a strong focus on liveness detection, BioID was able to present a beta version of the PhotoVerify technology as early as the beginning of 2018. PhotoVerify combines the above requirements by simply using two selfies for algorithm-based live detection and one ID photo snapshot taken with any camera for biometric comparison:
No need for strong bandwidth. No special hardware. No manual human interference. No cumbersome process.
PhotoVerify can be tested at the BioID Playground. For making a test integration of our Web-APIs into your services, please request a trial instance here: bwsportal.bioid.com/register.
Contact
Ann-Kathrin Schmitt
+49 911 9999 898 0
press@bioid.com