Identity Document Data Extraction & Authenticity Check

For users to have a more comfortable onboarding experience, our OCR performs identity document data extraction into predefined fields automatically.

The Optical Character Recognition (OCR) technology in our Digital Onboarding Toolkit allows for the easy guided capture of the ID and extraction of data, including the portrait image of the document holder.

During the onboarding process, the ID is first verified to make sure the user actually has the document.

Identity Document Data Extraction

Innovatrics AI-powered OCR technology works accurately and reliably in real time. We developed our OCR to support a wide range of alphabets, including traditional Chinese, Bengali, Cyrillic, and Arabic.

The OCR automatically captures and normalizes the photo of the ID, and detects the ID image for Know Your Customer (KYC) compliance.

Once the image of the document is captured, it is cropped and downscaled (on-device), and then sent to the OCR server for processing.
On the OCR server, all required fields of the document are parsed into text with an associated confidence score for each field.
The portrait image of the document holder is also extracted and converted into a secure proprietary biometric template for comparison.

See documentation

Remote Onboarding is easy with Biometric Solutions

Automated ID Classification

Once a picture of the ID is taken, our OCR automatically detects the document details, extracts the data from the ID and reads the MRZ zone.

With our technology, users mostly have to confirm the extracted data without the need for any corrections.

Automated ID Classifier identifies the document type, edition and issuing country from the captured picture.
Predefined fields of the onboarding form are automatically filled with the extracted personal data.

Try Demo

Document Annotation

Our OCR can read more than 60 types of identity documents.

Employing deep learning techniques and automating the majority of the process, it only takes a few days to train the OCR to read a new document.

To train a new document type, you only need to give it 4 pictures of a particular ID.
We train the neural network models specifically for each document to achieve the highest possible OCR accuracy.

Training new document biometric AI-system

MRZ, Barcode & NFC Readers

Our Digital Onboarding Toolkit also supports other technologies that can be used for data extraction.

Verify the data extracted from the ID by OCR using other validation checks.

Machine Readable Zone Reader and Parser works in real time and can run fully offline.
Barcode Reader and Decoder is compatible with the most common 1D and 2D barcodes.
You can use NFC in iOS and Android devices for extracting the data from chip IDs such as passports.

See documentation

Document Authenticity Validation

When onboarding new clients remotely, it is crucial to check if the document has not been forged or altered.

We cross-check dependencies between fields, which are read from the identity document to ensure its authenticity:

Cross-checking the data in MRZ versus data extracted by OCR and/or NFC (if used)
Field & picture authenticity – recognizing the over-stickered letters and pictures on the document
Color profile authenticity – checking the color profile of the document
Validating expiration date of the ID
Validating age and gender using biometrics
Verify photos taken during the onboarding process with the aid of biometrics

See documentation