Share via

Form Recognizer - Analyse Document in Label Data for Custom Model inconsistent results

Steffen 71 Reputation points
2023-04-06T11:36:54.7333333+00:00

Hello if I use the Form Recognizer Studio and want to Label Data to train a custom model, the analyzing of the uploaded PDF gives inconsistent or bad results. For Example it does not recognize text or it recognized the text but gives wrong values. For Example: Custom Model Label Data: User's image

Result from PreBuild Layout:User's image

If I use the Prebuilt Layout Model with the same PDF file, I get the expected results. As I understood the Documentation, the analyze function for the custom model Label Data uses the prebuilt Layout Model, why do I get difference results? Testet with the Preview API 2023-02-28 and the GA API 2022-08-31

Azure Document Intelligence in Foundry Tools
0 comments

Answer accepted by question author

VasaviLankipalle-MSFT 18,741 Reputation points Moderator
2023-04-06T19:15:39.31+00:00

Hi @Steffen , Thanks for using Microsoft Q&A Platform.

I have worked on similar issue earlier.

This is a model quality issue. Sorry for the inconveniences. The good news is that as recent updates (2023-02-28-Preview API) have improved the model, so you can expect better results in the next General Availability (GA), which is scheduled for this July. Please note that Custom models changes will take time until next GA.

However, you can still try pre-built models using latest 2023-02-28-preview API to check the updated results. I hope this helps.

Regards,

Vasavi

-Please kindly accept the answer and vote 'Yes' if you feel helpful to support the community, thanks.

Was this answer helpful?

0 additional answers

Sort by: Most helpful
Most helpful Newest Oldest

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.