At OCR Labs we pride ourselves in providing the most precise and intuitive ID Verification technology on the market. As part of this we’re focused on ensuring that our systems are accessible and inclusive regardless of physical disability, ethnicity, gender or socio-economic factors. Dan Aiello, one of our founders, explains how this journey towards inclusivity began and the results we’ve been able to achieve.
The importance of accessibility in digital identity verification
When we first developed our solution, accessibility was admittedly not top of our agenda. There are so many complex technical components to digital identity verification that it is easy for accessibility to become another checkbox on a long list of functional requirements. But then a bank, one of our very first customers, set up a meeting with their accessibility officer. The meeting was virtual, but as soon as he appeared on screen we could see he was completely blind. From that moment, accessibility became very personal to us, and we were determined to make it intrinsic to every aspect of our solution.
The initial feedback this accessibility officer gave us about the process flow and voiceover within our system was invaluable. We rewrote the code and worked with him to make sure the funnel is easier to navigate and that our compliance with WCAG accessibility standards is 100% comprehensive. From that early encounter, accessibility has been ingrained in everything we do. We take huge satisfaction knowing that everyone in society can access digital identity in the same way with our solution.
The transformational power of digital identity verification
We’re also aware that digital identity verification has enormous potential to transform people’s lives by democratising access to new opportunities and services. It can do this economically by giving them access to labour markets, for example. A 2021 McKinsey report found that by 2030, high adoption of digital ID will increase the growth of emerging economies by 6%. As much as 65% of that value could flow to individuals rather than to institutions.
Digital IDV can also unlock social value, opening up access to education and healthcare services. It can also improve civil rights and democracy by reducing voter fraud. There are many other benefits that promote social inclusion. In India, the right of residents to claim subsidised food through ration shops is protected because their identity and claim is authenticated through a remote digital ID system, rather than at the discretion of local officials. By providing greater legal protection, digital ID can also help eliminate child labour and enforce laws against child marriage.
We’ve developed many other technical measures to ensure these benefits reach all people worldwide.
Breaking down the technology barrier in digital identity verification
As part of this effort, we’ve made sure that people don’t necessarily need to have the latest technology to use our solution. Most of us in developed economies take our smartphones for granted. We can afford to upgrade our handsets to keep up with processing and data demands. But in many developing economies, and even in the poorer sections of western society, people tend to rely on recycled handsets that don’t have up-to-date processing power or operating systems. Many apps, whether they are Apple or Android, simply don’t work on these older phones.
To include the people who use older phones, we took the conscious decision to focus on making our solution available via browser on top of our iOS and android SDK. This means anyone with access to the internet can use our process flow. There is no need to download data-intensive apps or to have the latest operating system.
We’ve also looked at mobile networks, which can vary in speed and capacity across the world. A recent report from the GSMA stated that billions of mobile users in Africa, India and Latin America will remain on 2G and 3G up to 2025. That means they will have nowhere near the data capacity we are used to on 4G and 5G networks. Also, many people are on pay-as-you-go mobile data contracts and avoid any activity on their phones that is data intensive. To include these people, we make sure our solution continues to use the minimum amount of data possible. We want everyone who uses our solution to have the same frictionless user experience.
Overcoming language barriers in digital identity verification
We also believe that Digital IDV must be inclusive for people who can’t read, either because they haven’t access to the right education, or because they’re living in a different country and don’t speak the local language. To overcome this barrier, we have made our process flow very visual and found ways to remove as much text as possible. We also support over 40+ typesets and recognise all the relevant documents written in languages utilising those typesets.
Tackling the problem with racial bias in facial recognition
It may come as a rather shocking surprise to those that aren’t involved in this industry, but some Digital IDV solutions have problems recognising people who aren’t from that country’s majority ethnicity. We take this issue very seriously. When we were first developing our solution, we did consider using facial recognition technology from another supplier, but found that the error rate from all the suppliers we tried was unacceptably high. This error rate isn’t purely linked to race. A 2018 MIT study confirmed though that the error rate in facial recognition is consistently highest for people of colour, females, and the young. If you are all three of those, your chances of accessing digital identity verification are 30% lower than the rest of society.
One of the reasons for racial bias is that the algorithms that match components of the face use machine learning to improve the way they work. During development this machine learning is applied to real people, and those people tend to be the ones most accessible to the development team. In other words, the local population. If it is a western European company developing facial recognition, this means it will be better at recognising western Europeans rather than Asian or African people. Fortunately for us, OCR Labs is based in Australia, which has the highest proportion of immigrants among major western nations. So not only were we instinctively aware of the vast variety of skin tones, we were also able to test and develop the algorithms on the biggest range possible.
Another reason for the high error rate in most facial recognition is that the techniques they use are outdated. Many systems simply use one-dimensional measurements for the distances between parts of the face. For example, the focal point is your nose, your mouth area, your eyes, and so the system will essentially draw lines around those and then measure them. This leads to limited accuracy, and many queries that must be manually verified by real people. At best, this creates a delay of minutes, but potentially a delay of months if there is a major incident like a pandemic. At OCR Labs we have developed a very different approach. We use 3D techniques through our liveness testing. Combining these different elements enables us to have a robust, high accuracy version of facial recognition. Our technique also delivers a definitive result without the need for human intervention, which keeps verification fast.
An intuitive and inclusive customer experience
It’s important to us that everyone, no matter what their circumstance, can access the same process flow and has the same customer experience when using our solution.
We are also very aware that our solution may be the first interaction a customer has with a business. If that process takes too long or is hard to complete, the customer is likely to abandon it and choose another competitor. This means we do everything in our power to make the process as easy and quick as possible.
Our facial recognition does not require people to follow too many physical instructions or repeat words or numbers. Neither do we use extra lighting or flashing colours. All someone has to do is smile, just like they would in an everyday selfie.
We also don’t require the ID document to be in pristine condition and perfectly shot every time. If it’s at an angle, or slightly rotated, or if there is some wear and tear and the picture is low resolution, our proprietary technology will recognise it. As long as someone takes a picture of their ID with good intention, in everyday conditions, our system will be able to read it and identify it accurately.
Harmonious design for a harmonious world
At OCR Labs we believe in the potential of digital identity to positively impact people’s lives. We provide a solution that offers unrivalled inclusiveness regardless of people’s ethnicity or disability. In fact, we take user experience and product design as seriously as companies like Apple. To make this happen, we don’t buy in any third party technology and develop every part of our solution ourselves, including:
- Identity document recognition
- Document fraud assessment
- Liveness detection
- Video fraud assessment
- Face matching
Each component is built in synergy. The neural networks (or the ‘brains’) behind all our components are the same, so the interplay between them is seamless. At OCR Labs we call this harmonious design, and it has helped us create a digital identity verification solution that is accessible and inclusive for everyone in the world to use.
To learn more about how we can help you make identity verification more accessible and inclusive contact us via our demo form or drop us an email at firstname.lastname@example.org.