Azure/azure-sdk-for-python

Discrepancy Between Document Intelligence Python SDKs and UI Labels

SyedUmairHasan opened this issue · 1 comments

SyedUmairHasan commented
  • *azure-ai-formrecognizer and azure-ai-documentintelligence:
  • azure-ai-formrecognizer 3.3.3 and azure-ai-documentintelligence 1.0.0b2:
  • Windows:
  • 3.11:

Describe the bug
I trained a custom extraction model using Document Intelligence UI API version 2024-02-29-preview. When testing with both the Python SDK for Azure AI Document Intelligence and Azure AI Form Recognizer, the cell content of custom tables differs between the UI and the SDKs. The behaviour is not deterministic for some files.

To Reproduce
Steps to reproduce the behavior:

  1. Train custom extraction of model having complex tables
  2. Test from UI and python SDK

Expected behavior
The output of python SDK and UI should be same for all files

Screenshots

image
image
image

Additional context
I have created complete post with details in the MS QA forum:
https://learn.microsoft.com/en-us/answers/questions/1661837/discrepancy-between-document-intelligence-python-s?comment=question#newest-question-comment

github-actions commented

Thanks for the feedback! We are routing this to the appropriate team for follow-up. cc @ctstone @vkurpad.