Which SDK to Choose?

About our Product

The main product of our company is libraries written in C++.

Our libraries are adopted to any platform and architecture. Our product is environment-independent and is not bound to any OS version.

The C++ interface allows you to create wrappers for any popular programming language.

We provide wrappers for Java, C#, ObjC/Swift, Python3, PHP7/8 — for all your needs.

We can build an SDK package for the following OS families:

• RHEL, Debian, Alpine, Windows, MacOS, Android, iOS, WebAssembly;

There are certificates of compatibility with many operating systems:

• Astra, OS RED, ROSA "Chrome", Axiom Linux, etc.

Integration Types

Depending on where the recognition process is performed, the SDK can be classified into two categories: client-side and server-side.

Client-side Recognition

• Enhanced resistance to interference during recognition in video stream;
• Offline recognition is possible;
• Optimal user experience in non-biometric face comparison, two-way document recognition, and liveness check;
• Client's computing power is used for recognition.

What is recognition in video stream?

The system combines recognition results from different frames. This allows you to confidently recognize documents in conditions of poor lighting, glare, and other adverse factors.

Server-side recognition

• High speed of large images processing;
• No additional software and computation on the client-side is required.

Microservice

Microservice is a common case of implementing server-side recognition using our SDK.

Microservice is an approach in software development in which an application is built as a set of small, independent services, each of which performs a specific function and can be developed, deployed, and scaled independently.

Microservices can be deployed on physical or virtual machines. They can "listen" to local ports or interact with other services via an API. Local microservices can be configured to monitor the file system.

There is also a ready-made package for Linux/Windows in the form of a local server with a REST API, which takes minimum implementation time.

Desktop Application with a Graphical Interface

Advantages of a desktop application:

• Flexible interaction with hardware such as tablet scanners and webcams;
• Possibility of working offline;
• Maximum library performance.

Web Application

Advantages of a web application:

• Cross-platform;
• Accessibility from anywhere via the Internet;
• Posiibility to instantly update and deploy new features, that allows users to work with the latest version.

Our library can be integrated into the browser using WebAssembly technology. This approach enables execution of high-performance code directly in the browser. Working with web applications on mobile devices, you can easily interact with the camera and upload images through native interfaces.

Mobile Application

Advantages of a mobile application:

• Maximum library performance;
• Effective recognition in video stream;
• Optimal user experience in non-biometric face comparison, two-sided document recognition, and liveness check.

Integration Steps

Well, you need to recognize documents.

What steps should I take?

1. Choose functionality.
What do you need to recognize? What kind of documents, what kind of images? Scans, photos? Is it possible to use video stream?

2. Choose an integration type.
This can be using a script, running a service, microservice, or integration into your current solution. Each of the options differs in implementation time, operation time, and convenience of processing the result.

3. Define the response format.
Define the criteria for a 'correct response': which fields need to be recognized, how the data will be processed and stored.

4. Design in the “SE language”.
The “SE language” means using our interface elements.

5. Implement and test the solution.

How Long does the Integration Process Last?

Five minutes are enough to run the sample of our library supplied as part of the SDK. The sample will allow you to evaluate the operation of the recognition system — the speed, correctness, and completeness of the response. According to our customers' experience, these tests usually take one working day.

Embedding, writing, and testing the server-side use-case, in which the library returns only the document type and the contents of the text fields, takes about three to four days.

But for better understanding of the actual labor costs for building and implementing a recognition system that will meet your business needs, we recommend you to formulate the requirements for it in accordance with the task being solved. Understanding the structure of this system, you can write a technical specification for the recognition system. This technical specification will help you calculate the labor costs for integration.