Form Recognizer Studio Vs Sample Labelling Tools

Question

What is the difference between these 2 tools? Form Recognizer Studio and Sample Labeling Tool. Both can be use to create and train Custom Models.

I create a custom model using Form Recognizer Studio and Trained it , now how to I access it using python?

This is the script I'm using

import os
from azure.core.exceptions import ResourceNotFoundError
from azure.ai.formrecognizer import FormRecognizerClient
from azure.ai.formrecognizer import FormTrainingClient
from azure.core.credentials import AzureKeyCredential

endpoint= "URL"
credentials = AzureKeyCredential("key")

form_recognizer_client = FormRecognizerClient(endpoint, credentials)

form_training_client = FormTrainingClient(endpoint, credentials ))

custom_models = form_training_client.list_custom_models()

I am not getting back anything.

Answer 1

@Rajiv Ranjan Form recognizer studio is currently in preview and uses the form recognizer 3.x version of the API whereas the labeling tool uses 2.x version of the API.
3. x version of the form recognizer API will have some breaking changes so it will not be compliant with the labeling tool. However, you can still use the form recognizer studio with projects created in the labeling tool by creating a new project in the studio and pointing the storage connection to the blob container used with the labeling tool.

In any case, while using the SDK for python or any other language you will be able to list all custom models registered for your resource based on the endpoint and credentials used. If you are seeing no models listed above, you can cross check with the studio or just use the respective 2.x or 3.x REST API reference page to list the models using the subscription key of your resource. I hope this helps.

If an answer is helpful, please click on or upvote which might help other community members reading this thread.