Is it possible to extract titles, headers and table captions using Form Recognizer?

Question

I am using Form Recognizer to extract text and tables from multi-page PDF files. Is it possible to to extract the information whether a line is a title, header, subheader, ... and the caption of a table if a caption is present?

Answer 1

@Bogdan You can extract all the text from a PDF file but this text will not be grouped under the categories as mentioned above. Certain pre-built API's can extract details from a form or a ID card with a key/value pair in the response. The best option in this scenario is to use the custom form API to extract text based on training provided to a form. The training should essentially tag the text of a header or a caption with a name so that the actual endpoint with the custom model can extract the text as required. I would recommend to start using the form studio to train a model for your custom scenario.

If an answer is helpful, please click on or upvote which might help other community members reading this thread.