Sun Finance Uses Amazon Rekognition to Combat Identity Fraud and Remove Customer Friction

Initially, Sun Finance relied on a manual identity verification (IDV) process. However, many potential customers lack high-bandwidth internet connections and digital services such as email. To meet customers where they are, Sun Finance offers its lending solution on all mobile devices, including low-end devices with limited functionality.

This IDV process was time-consuming and resource intensive, requiring agents to manually check each applicant’s information and documentation. This time-intensive workflow was also prone to human error and inconsistencies.

In 2019, Sun Finance implemented Amazon Rekognition—a service that uses computer vision to extract information and insights from images and videos—to automate and streamline the IDV process. Using Amazon Rekognition, the company can accurately verify an applicant’s self-taken image (selfie) against their ID document, resulting in a smoother sign-up process.

“We receive all requests within 1 second,” says Sergei Kiriasov, head of risk technology at Sun Finance. “For each client, we request around 15 Amazon Rekognition operations, and they typically finish processing in 5–10 seconds.” The Amazon Rekognition latency is near real time, and the complete document verification process takes 15–20 seconds at maximum.

Sun Finance also implemented Amazon Textract—a service that automatically extracts text, handwriting, layout elements, and data from scanned documents—to help process its documents. “Amazon Textract is the only service we have tested so far that can accurately identify and read text written vertically or at an angle,” says Kiriasov.

Sun Finance uses several Amazon Rekognition features to accelerate its IDV process and detect duplicate or fraudulent account creation attempts. Face detection, for example, verifies that the user’s selfie is captured correctly. Face comparison measures the similarity between the user’s selfie and their identity document picture. Face index and search assist in creating a private repository of face vectors—mathematical representations of each face—that Sun Finance can use to determine the similarity of a face against other face vectors.

By automating IDV with facial recognition, Sun Finance not only enhanced its ability to mitigate risks associated with fraud and financial crimes but also improved the speed and accuracy of the application process. Now, customers enjoy a seamless onboarding process that makes it simple to apply for the financial services that they need.

“In some markets, the rate of automatically approved applications based mostly on Amazon Rekognition services reaches 60 percent,” says Kiriasov. “This is impressive considering the complexity of the registration process and the quality of the pictures we receive.”

In the near future, Sun Finance plans to integrate additional Amazon Rekognition features to enhance IDV, such as optical character recognition. The latest version of Face Liveness, which is currently being tested, helps verify that only real users—not bad actors using spoofs—can access its services.