The Center for Financial Technologies Basis (CFT Basis), which provides comprehensive digital solutions in mortgage lending, together with Smart Engines, has launched a service for automating the input of certificate data in the 2-NDFL form. The answer can significantly reduce the risk of erroneous data entry and speed up the processing of client documents by 2.5 times.
CFT “Basis,” in partnership with Smart Engines, announced the launch of an innovative service. Its peculiarity lies in using a unique synergy of automatic recognition algorithms and manual processing of documents, which ultimately guarantees the quality of the extracted information at the level of 99.9%.
Center for Financial Technologies “Basis” Smart Document Reader technology, developed by Smart Engines, is responsible for the mechanical part of the service. The verification and typing of documents are based on CFT BAZIS’ solution – Basis. Documents.
Information from the certificate in the form 2-NDFL (an official document on citizens’ income and tax deductions withheld from these incomes) is used by banks, developers, real estate agencies, and credit brokers to evaluate borrowers when applying for mortgage loans and tax deductions. The Smart Document Reader technology allows you to automate data entry from the 2-NDFL form on photographs and scans in mobile applications and Internet services. Center for Financial Technologies “Basis”
Smart Document Reader allows you to detect and extract up to several hundred attributes from the 2-NDFL certificate, including all tabular data, including when placed on two pages. The applied computationally efficient visual memory algorithms allow correcting projective distortions and achieving high recognition quality even in photographs taken by users in different lighting conditions.
The Basis. Documents service provides classification, verification, and verification of business logic in recognized data, which allows using document recognition technologies in the most complex business processes without the risk of errors and the involvement of additional human resources.
“The results of our scientific research in the field of computer vision and artificial intelligence provide high accuracy of recognition of certificate data in the 2-NDFL form in photographs, regardless of the lighting and device they were taken in. The use of ultra-fast neural networks and deep software and hardware optimization make it possible to recognize complex A4 documents in a few seconds even on devices that are modest in terms of performance,” comments Ph.D. Vladimir Arlazarov.
“The introduction of Smart Document Reader into the module of our Basis. Documents solution allows us to offer our customers a new experience using simple and secure data entry from 2-NDFL in mobile applications and Internet services. By automating data entry from a multi-line form of a tax certificate, we significantly reduce the time for processing an application, which is critical with a continuous flow of customers, and also reduce the percentage of erroneous data entry,” said Sergey Ponomarenko, CEO of CFT Basis.