AI will replace operators in the migration center of Moscow to check documents from foreigners

As TAdviser discovered, the Multifunctional Migration Center of Moscow (GBU Migration Center) decided to introduce a decision support system using a neural network. The organization placed the corresponding tender with an initial contract price ^[1] about 144 million rubles on the public procurement portal on August 28, 2025^[2]

The terms of reference indicate that such a system is needed, including in order to exclude the participation of the operator and conduct a fully automatic preliminary check of sets of electronic versions of documents (scan copies) that foreign citizens send to the information systems of the State Budgetary Institution Migration Center: "automation of the operator's work and its exclusion from the cycle of preliminary verification of electronic versions of documents."

𝓲

Фото: DilokaStudio/Freepik

We are talking about the sets of documents necessary to submit an application to the Migration Department of the Main Directorate of the Ministry of Internal Affairs for the city of Moscow and the Moscow Region in order to obtain a temporary residence permit in the Russian Federation, residence permit, citizenship of the Russian Federation, submit a notice of employment, undergo mandatory state fingerprint registration and photographing, medical examination.

In addition to excluding the operator from the process of preliminary verification of documents, among the goals of the project is to increase the availability of services provided by the GBU Migration Center, increase the number of foreign citizens receiving the services of the GBU Migration Center by automating the process of checking a set of documents, expand the range of services and services provided online and during a personal visit to this organization, etc.

It follows from the published technical assignment that the system will be able to recognize text information in graphic images of documents of foreign citizens established by law, analyze a separate document and automatically determine its type, conduct a comprehensive analysis of a whole set of documents, compare the data contained in one specific document with the data contained in the provided set of documents. After that, she will make a decision on the validity of the provided set of documents using neural network algorithms, and then provide the operator with the results of analyzing the set of documents of a foreign citizen.

For training neural networks, it is necessary to ensure, that is, collected and validated, an annotated sample at the rate of at least 8 thousand unique images per document. As a result, it is required to ensure the accuracy of recognition of text information in documents not less than 98%, indicated in the terms of reference.

The maximum number of documents that the system will be able to process per day at the time of project handover is 250 thousand, and the maximum number of external saved documents at the local storage by the time of project handover is set at 30 million.

The system databases should be managed using PostgreSQL 16 or DBMS similar in their characteristics.

The new system is planned to be built on a modular basis using a centralized three-level architecture for building information systems: data storage layer, business logic layer (application server) and presentation layer. The following environments are required in the process:

• Development environment, pilot operation environment and operating environment. It is used to develop algorithms computer vision and test on small samples (100-1000 pieces) of documents being developed. ON
• Pilot operation environment. Designed for testing software on large reference samples of documents (10 000 - 100 000 pieces). Also, this environment is intended for trial operation of software, for comparing in real time the results of the new document processing algorithm with the result of the algorithms in the operating environment (the processing speed and recognition accuracy are compared).
• Operating environment - an environment that uses a stable operating version of the software.

It is worth noting that digital technologies are increasingly used in the migration sector in Moscow. For example, from September 1, an experiment will also begin in Moscow and the Moscow Region on the mandatory registration of labor migrants from visa-free countries through the Amina application. Its use should replace the current migration accounting procedure. The application works by analogy with the "Social Monitoring" system, which was introduced during the coronavirus pandemic: with its help it will be possible to track the location of migrants.

See also

Digital Technologies in Migration

Electronic service for verification of documents of foreign citizens

Notes