Fine-tuning the Layout Extraction Model in Microsoft Document Intelligence

Question

Fine-tuning the Layout Extraction Model in Microsoft Document Intelligence

Stefan Raab 0

I came across the Layout Extraction model in Microsoft Document Intelligence, which has an accuracy of about 90% on my documents. I want to improve this accuracy further by fine-tuning the model on my own data. However, I cannot find a way to do this in the custom models. Can anyone suggest a way to fine-tune the Layout Extraction model in Microsoft Document Intelligence? Thank you in advance. -Stefan

santoshkc 15,355 Reputation points Microsoft External Staff Moderator

2024-02-08T10:14:16.9333333+00:00

Hi @Stefan Raab,

Following up to see if the below answer was helpful. If this answers your query, do click Accept Answer and Yes for was this answer helpful. Thank you.
santoshkc 15,355 Reputation points Microsoft External Staff Moderator

2024-02-08T10:50:10.51+00:00

Hi @Stefan Raab,

Thank you for your question.

Custom models are designed to extract data from forms and documents, as well as classify text. However, a single custom model cannot be used to both classify and extract text from a document.

Instead, you would need to create separate custom models for classification and extraction. For example, you could create a custom text classification model to classify the type of document, and then use a custom form extraction model to extract data from the classified document.

Thank you for your understanding!

2 answers

Your answer

santoshkc 15,355 Reputation points Microsoft External Staff Moderator

2024-02-08T10:14:16.9333333+00:00

Hi @Stefan Raab,

Following up to see if the below answer was helpful. If this answers your query, do click Accept Answer and Yes for was this answer helpful. Thank you.
santoshkc 15,355 Reputation points Microsoft External Staff Moderator

2024-02-08T10:50:10.51+00:00

Hi @Stefan Raab,

Thank you for your question.

Custom models are designed to extract data from forms and documents, as well as classify text. However, a single custom model cannot be used to both classify and extract text from a document.

Instead, you would need to create separate custom models for classification and extraction. For example, you could create a custom text classification model to classify the type of document, and then use a custom form extraction model to extract data from the classified document.

Thank you for your understanding!

Answer 1

santoshkc 15,355 Microsoft External Staff Moderator

Hi @Stefan Raab,

Thank you for reaching out to Microsoft Q&A forum!

The Layout Extraction model in Microsoft Document Intelligence is a pre-built mode, you can fine-tune the Layout model by Semantic chunking strategy. This is a new feature under preview state v4.0 (preview).

Also, you can try using the Custom model for your specific document types. The custom extraction model to analyze and extract fields from forms and documents specific to your business to create a document processing solution with more accuracy and performance.

Hope this helps. Thank you!

If this answers your query, do click Accept Answer and Yes for was this answer helpful.

Stefan Raab 0 Reputation points

2024-02-08T10:18:18.6966667+00:00

Hi @santoshkc , hmm, so actually, this doesn´t help too much in my case, but thank you for the answer!. The Semantic Chunking is a post-processing that we are already doing. And for the Custom Models, I don´t think that it is possible to use one label multiple times. For example, to classify and extract certain "paragraphs", "headlines", etc. Or am I missing out on something?

Answer 2

Stefan Raab 0

hi @santoshkc , I got an email that you have added a comment, but I cannot see the comment. You may have to submit it again. Thanks, Stefan

Share via

Fine-tuning the Layout Extraction Model in Microsoft Document Intelligence

2 answers

Your answer