CORE HR: OCR – Full Identity Field Coverage & Model Training

Overview

This release introduces an enhanced OCR capability powered by a custom-trained model to improve the extraction of information from identity documents. The previous OCR implementation relied on a predefined model that supported extraction of only a limited number of fields and struggled with documents that had varied layouts, formats, and complex structures.

The newly trained OCR model overcomes these limitations by enabling comprehensive field extraction from supported GCC identity documents and improving adaptability across diverse document templates and layouts.

Key Enhancements

1. Full Identity Field Coverage

• The OCR engine now extracts all required and configured identity document fields.
• Removes the limitations of the previous predefined field set.
• Supports dynamic identification and extraction of fields from identity documents.

2. Custom Model Training

• Implementation of a custom-trained OCR model designed specifically for supported GCC identity document types.
• Improved recognition accuracy for fields across different document layouts and formats.
• Better handling of complex, structured, and unstructured document formats.

3. Improved Data Mapping and Parsing

• Enhanced field mapping logic to ensure accurate alignment between extracted data and system fields.
• Increased reliability when processing documents with varied templates and layouts.

Impact
These improvements significantly enhance the reliability and usefulness of the OCR feature by ensuring complete and accurate extraction of identity document fields. Customers can now process supported GCC identity documents more efficiently, regardless of format variations or document complexity.

READ MORE