FAQ

The NV Global KYC platform allows companies to remotely verify user identities using document and facial recognition. It automates the KYC (Know Your Customer) process, combining document scanning, biometric verification, and other checks for fast and accurate identity confirmation.

The platform recognizes passports, national ID cards, driver's licenses, and other official documents from many countries. The specific list of supported documents is regularly updated – before integration, it is recommended to clarify whether the system covers documents from your country and of the required type. The list of supported documents is available at the link: [

If the document you need is not yet supported, you can contact NV Global support. The team will be able to assess the possibility of training the system on this document and release an update. Adding a new document type usually requires time for data collection and model retraining, but it is possible within the framework of cooperation.

It is recommended to photograph the document in good lighting, without flash and glare, so that all important areas (photo, stamps, numbers) are fully visible. Selfies should be taken with a good quality camera, in sufficient lighting, looking directly into the camera with a neutral facial expression – this will increase recognition accuracy and reduce the likelihood of errors.

The platform is by default oriented towards working with photos of original documents. If anti-fraud checks for photocopies and printouts (both black and white and color) are enabled, then such images may be recognized as suspicious and flagged with errors such as "image printed," "black and white image," or "image edited." However, if necessary, these checks can be disabled in the scenario editor. In this case, the system will accept high-quality copies and scanned documents (including printouts and Xerox copies) without issues, provided they meet the requirements for resolution, completeness, field visibility, and other parameters. To use copies, ensure that the corresponding anti-fraud rules are disabled in the verification scenario.

On NV Global servers, a full document check (including upload, recognition, field extraction, verification, and face matching) takes an average of 500–600 milliseconds, and in some cases, even less than half a second. This is achieved through an optimized architecture and high-performance neural networks. On the demonstration version servers (in demo mode), a similar check can take up to 5 seconds, which is due to limitations in computing resources and background loading. If the Liveness check (authenticity check) is included in the scenario, the total verification time may increase. In this case, the time for the user to pass the Liveness check ranges from a few seconds to one minute, depending on the shooting quality, lighting conditions, and the user's reaction speed.

A user can upload a document photo from a file, which is allowed by the platform. However, for the face verification stage (including Liveness), the use of a camera is mandatory. This is necessary to confirm liveness and exclude spoofing attempts – the system must ensure that it is a live person in front of it, not an image. Nevertheless, for photographing documents, we strongly recommend using the built-in camera component – WebSDK. This significantly reduces the risk of fraud. If anti-fraud checks within the current scenario are not critical, and the scenario allows uploading files without protection against falsification, then the user can provide document images directly from the gallery or by uploading a file.

The system accepts images exclusively in JPEG or PNG formats. Files in other formats (e.g., HEIC, WEBP, PDF, etc.) will be rejected. Image size requirements apply not to the entire file, but to the size of the object in the image: The document in the photo must occupy at least 200 pixels on its larger side (width or height). If it is smaller, the system will return the error "too small image" or "document too small" because it will not be able to extract data correctly.

The user's face (e.g., in a selfie or document) must be at least 30×30 pixels. Smaller faces cannot be analyzed and recognized properly and will lead to the error "face too small".

It is also important to avoid strong compression, blurriness, and poor lighting – such defects can degrade OCR quality and reduce the accuracy of face comparison.

NV Global can automatically recognize which document was uploaded (e.g., passport or driver's license, and from which country, if the format is unique). You don't need to manually select the type – just upload the photo, and the algorithm will classify the document itself. However, in some cases (if documents of different types are very similar), the system may require specifying the type in the verification scheme settings.

If the system incorrectly recognized the text on the document, it is likely that the image quality was insufficient (glare, blurriness). In such a situation, it is better to take a new photo of the document with better quality and try again – this most often solves the problem. If the recognition error repeats on a clear image, you can report it to support, and in the meantime, manually correct this data outside the system.

The NV Global algorithm analyzes image quality and the presence of mandatory elements (stamp, signature, owner's photo) and can detect copies or fakes. However, a full authentication of security holograms or UV markings from a regular photograph is difficult. Thus, the system identifies obvious discrepancies (document format, fake elements) but does not guarantee the detection of all types of fakes, especially if they are not reflected in the photo.

If the identity document consists of two pages (like an internal passport with registration), the platform can process both pages sequentially. In the schema settings, you can specify the need for an additional step – first, a scan of the main page, then, for example, the registration page. The user will be prompted to upload both sides, and the system will combine this data for a complete verification.

NV Global can read text in Russian and other languages with the Cyrillic alphabet, as well as in Latin script (English, French, etc.), and in many other languages represented in documents. The system is trained on international samples, so it automatically extracts fields (full name, date of birth, document number, etc.) regardless of the language – the only important thing is that the document is on the list of supported formats.

Enface platform's Liveness check is used to confirm that a real living person is in front of the camera, not a photo, video, or other deception. We use three modes of active Liveness check: — Simple tasks — the user performs individual commands (e.g., turn head left or right, blink, look at a specific point). — Complex tasks — tasks are complicated and follow in quick succession (e.g., combinations of turns and gazes) for deeper verification. — Circular head movement — the user moves their head in a circle, similar to the initial Face ID setup on an iPhone. This method allows for collecting the maximum amount of 3D facial information.

During task execution, the system takes a series of snapshots (key frames) which are automatically sent to the server. On the server, these images are analyzed using neural network models that determine the presence of signs of a living person (micro-movements, skin texture, depth, etc.). Based on the analysis of all frames, the system makes a final decision on passing the Liveness check.

If the liveness check fails, the system could not verify that it was a live person (e.g., due to poor lighting or failure to perform actions). It is recommended to ask the user to repeat the liveness check: look directly into the camera, do not cover your face, and perform the required action. The system usually allows several attempts. If it fails repeatedly, it may be possible to switch to an alternative identification method (e.g., manual verification by an operator).

The Liveness test is recommended to ensure that the document is presented by its owner at the time of verification. Nevertheless, the platform is flexible – at the client's request, a simplified KYC scheme can be used, which includes only document upload and recognition. In this case, the user's face is not confirmed via selfie, which somewhat increases the risks of fraud, so disabling Liveness makes sense only in low-risk scenarios.

Built-in Liveness mechanisms check for natural behavior: they can distinguish a live face from a static photograph or a video playback. If, for example, an attacker wears a silicone mask or tries to show another person's video on the camera screen, the system will very likely recognize this as an attempted deception and will not count the liveness check.

NV Global implements a number of anti-fraud mechanisms: it checks the document for forgery (for example, by recognizing a copy or photomontage), compares facial biometrics with those already known in the system (to identify if the same person has tried to register under a different name), and also monitors the sequence of user actions. Any deviations (black and white copy of the document, mismatch between the face and data, suspiciously frequent attempts, etc.) are flagged and can serve as a signal for rejection or additional verification.

The label [fraud] image printed indicates the detection of signs of a copy: the system determined that the document is likely not an original, but a scan or photocopy (e.g., black and white or color). This error serves as a signal of possible fraud. It is recommended to request a photo of the original document from the user – with real security features – to pass the verification.

The platform supports the storage of biometric facial templates of users who have successfully passed verification. During new KYC attempts, the system automatically compares the face with the database of already known embeddings.

If the face has been encountered in previous sessions—even under a different name, with a different document, or in a different context—the system will detect a match. This allows for timely identification of repeated KYC attempts with data substitution.

In the interface, upon a match, the system highlights the current session ID in red, which is a direct indicator that the person has already appeared in the system. Clicking on this ID opens the history of all sessions in which this face was detected, with the ability to view and analyze them.

The platform supports a built-in mechanism for adding faces to a blacklist. This allows for the automatic blocking of verification attempts by the same person, even if they use different documents or personal data.

To do this, click on the icon with a crossed-out circle in the session viewing interface. After that, the face detected in this session will be added to the blacklist. During subsequent verification attempts:

• If the face matches a previously blocked one, the session will be automatically rejected.
• This will be accompanied by the label "User on blacklist" in the interface.

Blacklist in the interface

This approach is effective against repeated fraudulent attempts with data substitution, as it relies on biometrics rather than just document details. The mechanism does not require additional implementation from the client – it works out of the box.

All technical documentation is available on the NV Global portal: it describes REST API endpoints, request and response formats, examples, and errors. For ease of integration, you can use Swagger/UI at or read the guide on GitBook. It is advisable to familiarize yourself with these materials before starting the integration.

To gain access to the platform's API, you need to:

Request a unique promo code from the NV Global administrator.

Register on the platform (via the web interface) using this promo code.

After registration, you will get access to your personal account, where you can:

    Manage connections,

    Obtain tokens,

    Configure scenarios and integration,

    View statistics and request status.

All access parameters (tokens, identifiers, configurations) are issued automatically through the platform – we do not transfer anything manually, including any clientKey or secret keys.

Usually, a Non-Disclosure Agreement (NDA) is concluded before this. After that, you can start using the test access. Load parameters, limits, and scenario restrictions are discussed individually.

After a full contract is signed, production access will be granted, with the possibility of connecting to an isolated environment, dedicated servers, and custom SLA parameters.

The general process is as follows: your backend calls the NV Global API method, passing the document image and the user's selfie/video (either in sequential requests or within a single session, depending on the chosen scheme). The NV Global server processes the data (OCR, face comparison, liveness) and returns a structure with the results – the verification status and extracted data. In the integration, you will need to implement sending the data, receiving the response, and parsing the results (e.g., filling out a user form with data from the passport, processing the "success/failure" status).

NV Global is developing a Digital ID solution, within which a record (digital passport) is created in a distributed system (blockchain) based on KYC results.

• What is a digital passport on the blockchain:
• A unified verified identity profile created after successful KYC completion;
• Stored in a distributed blockchain network — decentralized and immutable;
• Allows for the reuse of verified data across different services without the need for re-identification;
• Can be implemented as a Soulbound NFT token or another digital certificate;
• Recognized by ecosystem partners, facilitating access to various services.

Essentially, it's an add-on to KYC: undergo identification once — receive a digital certificate that can be presented repeatedly.