New features in the Azure Form Recognizer client libraries

• Build a custom classification model for document splitting and classification: The Form Recognizer client libraries now allow users to build a custom classification model for analyzing input documents. This model can be trained to split and classify documents based on specific criteria, such as loan application packages containing application forms, payslips, and bank statements.
• Add-on recognition capabilities: Users can now enable additional analysis capabilities in the Form Recognizer library. These include recognizing font-related properties of extracted text, detecting barcodes in documents, and detecting formulas in scientific document types.
• Barcode recognition: The Form Recognizer library now includes a feature for detecting barcodes in documents. This feature provides the decoded value of the barcode, as well as an overall extraction confidence.
• High-resolution recognition: Users can now enable the high-resolution recognition add-on feature when analyzing documents. This feature allows for more accurate extraction of text and other data from documents.
• Detect formulas: The Form Recognizer library can now detect formulas in scientific document types. It provides information about the type of formula (inline or display) and its LaTeX representation, along with its polygon coordinates.
• Font extraction: The Form Recognizer library now includes a feature for extracting font properties associated with extracted text in documents. This feature provides information about the font family, size, weight, style, and color of the text.

The new Form Recognizer client libraries are now available for download from each language's preferred package manager. Users can find example code for building a custom classification model and analyzing documents using this model in Java, .NET, Python, and JavaScript/TypeScript.