Compose Custom Extraction and Custom Classification Models

Darius Irani 0

After reading the documentation on composing custom models, I'm still confused on how to implement it.

I have a few trained custom extraction model's I'd like to compose with a custom classification model - I would like the classification model to first sort the type of document, and then send it to the correct custom extraction model to extract the correct fields.

If this is possible, please let me know, and please advise me on the steps to achieve this. Thanks.

santoshkc 6,955 Reputation points Microsoft Vendor

2023-10-04T13:26:11.7533333+00:00
Hi @Darius Irani ,

Thank you for reaching out to Microsoft Q&A! I will be happy to help you with your query.

Build and train a custom classification model:

Start by navigating to the Document Intelligence Studio.

In the Studio, select the Custom classification model tile, on the custom models section of the page and select the Create a project button.

1. On the create project dialog, provide a name for your project, optionally a description, and select continue. 2. Next, choose or create a Document Intelligence resource before you select continue.

3. Next select the storage account you used to upload your custom model training dataset. The Folder path should be empty if your training documents are in the root of the container. If your documents are in a subfolder, enter the relative path from the container root in the Folder path field. Once your storage account is configured, select continue.

4. Training a custom classifier requires the output from the Layout model for each document in your dataset. Run layout on all documents prior to the model training process. 5. Finally, review your project settings and select Create Project to create a new project. You should now be in the labeling window and see the files in your dataset listed. Label your data: In your project, you only need to label each document with the appropriate class label. Train your model: With your dataset labeled, you're now ready to train your model. Select the train button in the upper-right corner. Test the model: Once the model training is complete, you can test your model by selecting the model on the models list page. Watch the tutorial: Discover how to navigate through the new custom classification labeling process. Form Recognizer Studio - Microsoft Azure

Regarding your query about sending the Custom Classification Model output to correct custom extraction model to extract the correct fields: Use the Language Studio API to classify a document(classification model endpoint URL) and extract fields(custom extraction model endpoint URL) A classification model can be paired with a custom extraction model to analyze and extract fields from forms and documents specific to your business to create a document processing solution. Standalone custom extraction models can be combined to create composed models.

For more info: Custom document models - Document Intelligence (formerly Form Recognizer) - Azure AI services | Microsoft Learn

Hope this helps. Thank you.
santoshkc 6,955 Reputation points Microsoft Vendor

2023-10-05T12:52:38.4966667+00:00

Hi @Darius Irani ,

Following up to see if the above suggestion was helpful. Thank you
santoshkc 6,955 Reputation points Microsoft Vendor

2023-10-06T12:56:49.2433333+00:00

Hi @Darius Irani ,

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution, please do share that same with the community as it can be helpful to others. Thank you.

Share via

Compose Custom Extraction and Custom Classification Models