Accessibility & Inclusivity at OCR Labs
Effective starting: October 2021
Everyone should be able to verify themselves and access any remote service from wherever they are. We’re committed to making that a reality.
When we first started building our identity verification solution, we saw accessibility as a checkbox exercise. It was only when we met with the Director of Accessibility at a large bank here in Australia, and saw that he was blind, that we understood what it meant to build an inclusive product. Since that day, we’ve kept accessibility and inclusivity front of mind. Here’s how we’re making identity verification accessible for everyone.
Visual and Audio Accessibility
Text Size - For people with impaired vision, the text on a page should respect user-specified defaults or be dynamically resizable to be legible. OCR Labs conducts verification using a user’s installed mobile browser. This means that the user can easily change the base font size in their phone settings, so that all text scales appropriately.
Language - We’ve translated our onboarding flow into 20+ languages with 46 typesets too. We also support the voiceover functionality within your smartphone, which a user can activate, and follow the instructions to complete verification.
Colour Blindness - There are approximately 300 million people around the world with some form of colour blindness. Most colour blind people are able to see objects clearly, but are unable to fully 'see' red, green or blue light.
We follow the WCAG best practice guidelines by ensuring that:
- All text and interactive elements have a color contrast ratio of at least 4.5:1.
- Colour is not the only indicator for interactive elements within our workflow. We also underline interactive elements.
- While green is one of our brand colours, green and red are the most common colours for colour blindness. We avoid using red or green colours in the user verification instructions or workflow, so that users who cannot see the full color spectrum are able to read all text.
Identifying and Onboarding Every User, Every Skin Tone
Algorithms don’t always see all faces equally. Studies conducted by the National Institute of Standards and Technology have found that the algorithms have a harder time recognizing people with darker skin.
Facial recognition technology has been around for decades. But it primarily focused on whether an image contains a face, and whether it matches the face in another image. To do this, most facial recognition systems measure the distances between certain facial features, like the space between the eyes for example. Skin tone has traditionally not been considered at all. As a result, facial recognition can be less accurate for different ethnicities.
We’ve built our facial recognition technology ourselves, and skin tone is one of the primary factors we consider when matching a selfie with a user’s identity document. Combined with a live photo, which considers three dimensions rather than just two, we are able to take into skin tone into consideration.
Building a solution that ensures the user is live, and their selfie isn’t just a photo of a photo is a challenge for identity verification providers. Other providers flash lights off the user’s face or request that they turn their head or wave. Both these liveness proof solutions can create challenges, such as requiring movements which disabled users might not be able to do.
Like Apple’s live photo, OCR Labs captures the feed around the photo being taken and analyses it for liveness. We use technology including analysing light refraction from the face and background objects to determine a user’s liveness.
Overcoming Technology & Bandwidth Barriers
Smartphone Access - Smartphone use is not equal everywhere. Half of non-users around the world say that the high cost of purchasing a smartphone is the reason for not owning one. For those that do own or share a smartphone, it often is an older model with lower quality cameras.
We’ve built our identity verification solution to utilise images from extreme scenarios. We normalise user photos to account for variability with lighting levels and lower quality cameras.
Bandwidth - In some countries, mobile data is expensive and/or slow. Identity verification can be a barrier to access as it requires the user to take photos which are then sent across networks.
Other identity verification providers embed their verification workflow into the app winto the workflow of their customer’s app as an SDK. Not only does this increase the size of an app, which increases the cost of downloading it, but it usually means the SDK requires more bandwidth to transmit the user’s selfie and identity document photo to be verified. We embed our workflow on a browser which means that the app is smaller, and customer’s can verify themselves even if they are in a 2G zone.
Building a Diverse and Inclusive Workforce
At OCR Labs we believe in the inherent strengths of a vibrant, diverse and inclusive workforce. Our different backgrounds, perspectives and life experiences assist OCR Labs in building lasting connections with our people and clients. We believe in empowering the potential in everyone.
Accessibility is part of who we are. We are proud to be an equal opportunity employer and we value diversity. We do not discriminate on the basis of educational attainment, race, religion, colour, national origin, gender, sexual orientation, age, marital status, veteran status or disability status - simply, we consider all qualified applicants, consistent with any legal requirements. If you have a disability or special need that requires accommodation, please let us know. You can view our open roles here.
Contact: If you have questions or concerns, contact us at firstname.lastname@example.org