Smart Engines has announced a new version of its optical document recognition system. Based on the proprietary AI platform HIEROGLYPH, the system is able to recognize more than 900 document types in more than 40 languages from 180+ countries. In the latest SDK version the platform deploys an advanced 8-bit OCR technology, as well as latest stage of algorithms for identification and orientation of documents.
Smart Engines CEO Vladimir Arlazarov, Ph.D, stated: “In 2015, we introduced the then-first solution that was able to recognize Russian passport on mobile devices in the video stream and in single images. Since then, our scientific achievements in the field of computer vision have allowed us to come up with a market offering that can transform the user experience in terms of document processing in images and in a video stream in real time. We are talking fast algorithms for restoring the geometry of documents on images with projective distortions, ensuring highest level of accuracy in the recognition results, regardless of lighting conditions and the quality of the device on which the recognition is performed.”
With the release of the new version, the users can now obtain the automatic verification of document validity evidences. The system verifies the recognized textual data of the document’s visual inspection zone (VIZ) and the machine readable zone (MRZ), if available. To perform the document validity check, the system compares the current date with the date of birth, the document issue date and its validity period.
Version 3.0 supports the recognition of 940 types of documents, including passports from all countries of the world, identity cards (ID cards), driver’s licenses, visas, health insurance cards, etc. The software recognizes passports, ID cards and driver’s licenses of every European and CIS country, and got the list of supported documents expanded for such countries as Australia, Canada, China, South Korea, Mexico, USA and Japan.
The new version advances the recognition of all types of bank cards, machine-readable zones (MRZ) on any document, and one-dimensional / two-dimensional bar codes. In the case of MRZ, this is due to an improved text string search algorithm. The main efforts in the development of barcode recognition technology were aimed at improving the quality of data extraction from AZTECп and QR, which are widely used on utility bills, transport tickets, and other generic documents for commercial tracking and validity verification.
Furthermore, version 3.0 supports the recognition of bank cards not only in the video stream but also in single images. This feature provides better service for payment solutions companies, as well as those associated with mobile banking, as it allows the customer to make a payment by using an image from the gallery, without activating the smartphone’s camera.
“One of our main principles is the autonomy of our technology. During the recognition process, all calculations are always performed on the customer’s device. That can be either a regular smartphone (in cases when the Smart IDReader is built directly into a mobile application), or desktop computer, as well as server or self-service terminal. In any case, we offer the so-called “in-house” processing, where our users do not transfer information about their clients to external services and can thus implement high standards of personal data security, ” Vladimir Arlazarov added.
Currently, Smart Engines serves a broad and diverse client base, which includes more than 35 banks and insurance companies, 15 system integrators, 18 service providers, leading telecom operators, 6 ticket-selling companies. The technology is also incorporated in various projects with remote and on-site service. Among Smart Engines clients are: Tessi, S&T, Dukascopy Bank, Tinkoff, Alfa-Bank, Post Bank, Renaissance Insurance, BIG2 Russian telecoms — MTS and MegaFon, SkyTeam airline — Aeroflot, and more.
Smart Engines SDK algorithmic stack is optimized for x86/x86_64, ARM, SPARC, MIPS, and Elbrus architectures and is compatible with such operating systems as iOS, Android, Windows, Linux-based systems, Avrora MOS (formerly Sailfish), and macOS.