Smart Document Engine uses state-of-the-art AI methods to automatically detect and classify documents in the image stream, detect and recognize tables, labels, marks and text details. Automatic document classification and recognition are robust to both various distortions of mobile phone and scanning devices, and document layouts. Smart Engines specialists developed the latest generation of GreenOCR® optical character recognition technology to ensure superior text recognition quality and speed. Smart Document Engine precisely and quickly recognizes payment documents, invoices, KYC questionnaires, consents forms, such as consent to personal data processing, or proxy forms both from scanned documents captured using ordinary office or specialized sheetfed scanners, and from photos captured on document cameras or regular mobile phones and tablets.
Smart Document Engine is a full-fledged multipurpose system for recognition of documents with flexible and fixed layouts. The system provides automatic data entry and verification of various questionnaires and documents with high speed and precision. The Smart Document Engine solution can be applied for automatic data entry of banking documents, KYC questionnaires and other documents both in offices and service points, as well as in the context of remote customer service. The solution provides data recognition of proxy forms, consent forms and other structured documents for onboarding purposes. The Smart Document Engine form type classification module allows you to implement completeness verification when entering and registering a package of documents. Automatic recognition questionnaires with complex structure allows you to simplify your onboarding process — insurance, banking, financial, or legal services; personal data entry when performing medical tests; exam and test data processing; submitting applications for public services, and many other processes. Subsystems for tabular data analysis, marks (such as checkboxes) detection and recognition, as well as analysis of handwritten notes, allow you to automatically enter and recognize documents with a complex structure and perform cross-verification of data.
Thanks to the state-of-the-art methods of computer vision, machine learning, AI and careful optimization, the Smart Document Engine SDK allows to integrate the feature of scanning and recognizing payment documents, invoices, KYC questionnaires, consent and proxy forms in any application for mobile phones and tablets, office applications, integration buses, and data entry centers. The SDK does not heavily consume resources: payment order scanning and recognition on a low-end mobile phone takes about 2-3 seconds. Thus, a mobile phone turns into a high-performance document scanner. The requirements for mobile device cameras are modest as well: a regular 8-megapixel camera with autofocus is enough. The system performs automatic document detection, classification, and then an automatic location analysis of the details and tables, OCR of typed fields, automatic ICR of hand-printed fields, OMR for labels and automatic 1D and 2D barcode recognition. The Smart Document Engine solution also provides a built-in tabular data processing module, as well as an extraction and verification module for graphic elements (such as stamps and signatures).
The Smart Document Engine system is an independent on-premises solution and does not transfer images to any third-party servers for processing and recognition, thus, sensitive data transfer to third parties is excluded. Consequently, Smart Document Engine is the most secure solution for both business and its customers. The Smart Document Engine SDK does NOT transfer data and images for processing to Smart Engines or to any third-party services, does NOT save data or images (the processing is carried out in the device’s local RAM) and does NOT require internet access. The Smart Document Engine SDK is GDPR, CCPA, HIPAA-compliant.
Smart Document Engine is delivered as an autonomous SDK (software development kit) to integrate it into the customer’s infrastructure and applications. The SDK includes OCR libraries compiled and optimized for a set of target platforms and operating systems, API documentation (application programming interface) and integration examples for various programming languages. The scanning and recognition API for questionnaires, forms, and other documents recognition allows you to integrate this functionality into programs and applications built using C, C++, C#, Java, Python, PHP, Objective C and Swift. Optionally, the SDK can be bundled with the Smart ID Engine and Smart Code Engine systems to scan and recognize ID documents and codified objects using a unified software interaction interface.
Thanks to a simple API, the Smart Document Engine SDK can turn iOS or Android OS phones into a scanner of bank documents, invoices, proxy forms, certificates, and other documents with fixed and flexible layouts. Our proprietary AI and computer vision algorithms provide high-speed (about 2 seconds per recognition of a full-size photo on iPhone XR) and autonomous document recognition not only in English, but also in other languages of the world.
Deep software and hardware optimization of the code and algorithmic recognition principles for various processor architectures allows Smart Document Engine to autonomously recognize bank documents, invoices, and other documents with fixed and flexible layouts on any personal computer (including nettops and all-in-one PCs), and turn them into high-performance workstations for entering and processing document images captured from various image capture devices (flatbed and sheetfed scanners, specialized document scanners and document cameras). A wide range of software interfaces provides the possibility of native integration into most information systems.
Thanks to thread safety and efficient resources management, the Smart Document Engine SDK allows you to implement scalable AI platforms for image processing of payment documents and invoices, certificates, consent and proxy forms, as well as other documents with fixed and flexible layouts. Unique algorithms for automatic document classification and localization in images (scans and photos) allow you to automatically process heterogeneous streams of input images. The recognition performance of complex data structured documents on modern computers (with an AMD Threadripper 3970X processor) is about 10 full-size photos per second, which allows our customers to create highly secure and efficient WebAPI services and integrate the SDK into existing data exchange buses.
Please fill out the form to get more information about the products,
pricing and trial SDK for Android, iOS, Linux, Windows.