Can I Perform High-Resolution Analysis for PDFs in Document Intelligence Studio?

Josias Hartmann 0

Hey, I hope you're all doing fine! I'm having an issue with the custom extraction model in Document Intelligence Studio. I'm trying to extract information from a very clear PDF, but when I run the layout to start labeling items, it doesn't seem to identify clearly visible content. Is there a way to perform a high-resolution analysis similar to the 'layout' document analysis?

VasaviLankipalle-MSFT 17,121 Reputation points

2024-07-09T02:58:30.0433333+00:00

Hello @Josias Hartmann , Thanks for using Microsoft Q&A Platform.

May I know thew API version you are working on? Do you mean that some information is not available for labeling?

Is it possible to share the document or display the results in a private message as requested?
Josias Hartmann 0 Reputation points

2024-07-09T13:59:20.8833333+00:00

Hey @VasaviLankipalle-MSFT !

I'm following the steps in the instructional videos on the "Custom Extraction Models" in the Document Intelligence Studio GUI. The API version I'm working with is "2024-02-29-preview".

Let me try to explain a little better.

When I run "Layout" in the "Custom Extraction Model" (https://documentintelligence.ai.azure.com/studio/custommodel/), there are some fields that the model apparently can't find (see the image here). However, when I run a document analysis in https://documentintelligence.ai.azure.com/studio/layout with premium detection and high resolution checked, everything is fine (see it here).

So, I was wondering if I could use premium detection when running the layout in the custom extraction model.

I hope I made myself clear. Thanks for your help!
Josias Hartmann 0 Reputation points

2024-07-11T14:29:26.62+00:00

Hey @VasaviLankipalle-MSFT

Thanks for getting back to me so quickly.

I've tried what you suggested, but I'm still having trouble with some fields not appearing after a regular analysis, as I showed you before. Do you have any other tricks to improve accuracy? Maybe there's something I'm missing. I'd really appreciate any ideas you might have.

Thanks again for your help!

1 answer

VasaviLankipalle-MSFT 17,121 Reputation points

2024-07-10T23:04:30.3766667+00:00
Hello @Josias Hartmann , thanks for providing detailed information.

Unfortunately, premium detection in the custom extraction model is not available.

In this case, I would suggest you to auto-label the table fields.

Firstly, add the "Table" field to define the field.

After "Run layout" is done, a "table" mark will show up. Click on the icon/mark.

A table with the recognized content will show up. Scroll to the bottom and click "Auto label" and choose the table field accordingly.

A "Table labeling" wizard will show up in the right side and you can edit and review the rows/columns accordingly.
Please sign in to rate this answer.

1 person found this answer helpful.

0 comments No comments
Sign in to comment

Use comments to ask for clarification, additional information, or improvements to the question.

Share via

Can I Perform High-Resolution Analysis for PDFs in Document Intelligence Studio?

1 answer

Your answer