Karza's AI-powered Video KYC Enables End-to-End Automation in Onboarding
By: Karza Technologies
The Reserve Bank of India amended its master directions on know-your-customer norms on May 12 2021 to encourage video-based customer identification and simplify periodic KYC updations for bank customers. Video-based customer identification, also known as Video-KYC, can be used for new account openings and for periodic updates of KYC for existing bank customers.
In the midst of the lockdown, Video-KYC became the need of the hour to onboard new customers, who were confined to their homes but were still willing to open their bank accounts to facilitate digital transactions. The use of Video-KYC was further expedited and proliferated with RBI allowing the use of Video KYC banking entities to convert limited accounts to complete KYC-compliant accounts.
During that phase Karza Technologies introduced their Video KYC solution. As of today, Karza's Video KYC solution is used by more than 30 regulated entities, and the average call lasts less than 180 seconds, which has resulted in humongous cost savings. It is 100% RBI-compliant, both in terms of infrastructure and processes, and enables end-to-end onboarding automation.
One of the most unique highlights of Video KYC by Karza Technologies is that it caters to India's specific needs. In designing their cognitive algorithms, Karza considered the possibility of low resolution, Indian faces, worn documents, shadowed/blurred or skewed images, reducing false positives by matching the photos contained in the documents with the image of the face captured during the live video call.
As part of their proprietary models and algorithms designed for Indian names, the Video KYC solution employs features such as space predictors for name strings, gender classifiers, name scores, and surname scores.
Another factor of consideration is that in India, addresses can take many forms, so Karza's solution compartmentalizes addresses into different buckets and compares the consolidated addresses with efficient fuzzy retrieval from locality databases.
During the face match process in addition to handling rotated faces, the algorithm also takes into account India-specific face attribute models. Karza verifies liveness with 100% accuracy.
To add to it Karza's OCR models are template-agnostic and work with a minimalistic set of requirements. Due to the fact that OCR requirements are often subject to sudden regulatory changes, this agility makes it ideal for applications. In addition to being able to train models quickly, Karza's OCR solution offers acceptable accuracy even when poor quality images are provided.
More Info: https://karza.in/