Problems to extract data with custom model

Question

I am training a custom model in form recognizer to process pdfs with 50 pages each, each page has similar information and tables.

However, when starting the analysis, it only extracts the data from the first page. When trying to label the fields of the following pages with the labels that I already created, it tells me that I cannot use the same labels on several sheets. Is it possible that I am labeling incorrectly? Or is it necessary to create different labels for each sheet?

I leave an image as an example of the design of each page of the pdf that I try to extract.

Thank you.

Answer 1

ricardoyepez For the scenario where you are seeing the message about using multiple labels across pages, this is a valid message where the labels cannot span multiple pages. You need to add a new label with a different name. For example: For a piece of text that occurs at similar region across pages, try to suffix a page number or counter like label-1, label-2, etc..

This can help you build a labeling project for all the pages without compromising on the result of the analyze operation.

Another solution that can help is to pre-process the form if similar information is in each of the page you can split the form or document in 50 different form and pass each form to the API thereby reducing the effort to create a model by reducing the labeling activity of a project.

You could also try any of the pre-built models if the form is consistent and the language is supported for that pre-built model.

If this answers your query, do click Accept Answer and Yes for was this answer helpful. And, if you have any further query do let us know.