While Azure offers prebuilt receipts and invoices in the Form Recognizer service, you need to extract data from the forms specific to your business.
Hence, you decide to use custom models of Form Recognizer in Azure.
You further utilize these custom models into a custom skill interface of your enrichment pipeline.
Given the steps below, sequence them in the correct order of execution to create, use and manage your custom model in Form Recognizer: Step 1: Manage your custom models. Step 2: Use the Azure Blob container to upload training data. Step 3: Establish your training dataset. Step 4: Use labeled or unlabeled datasets to train the model. Step 5: Test the model using a dataset not used in training. Step 6: Create a custom skill and use it in the enrichment pipeline.
Click on the arrows to vote for the correct answer
A. B. C. D.Correct Answer: B.
Option A is incorrect because you will manage your model once it is created, trained and tested.
You would also need to upload at least five forms for the training dataset before proceeding with the next steps in the execution process.
Option B is correct.
Here is the correct order of executions steps:
Step 1: Establish your training dataset.
Step 2: Use the Azure Blob container to upload training data.
Step 3: Use labeled or unlabeled datasets to train the model.
Step 4: Test the model using a dataset not used in training.
Step 5: Manage your custom models.
Once you have created the custom model in Form Recognizer, you can create a custom skill and use it in the AI enrichment pipeline of Cognitive Search solution to analyze documents.
Option C is incorrect because you will manage your model once it is created, trained and tested.
Option D is incorrect because you will manage your model once it is created, trained and tested.
Reference:
To learn more about building and optimize a custom model for Form Recognizer, use the link given below:
The correct sequence of steps to create, use and manage a custom model in Form Recognizer in Azure, and utilize it in a custom skill interface of your enrichment pipeline, is as follows:
Step 1: Manage your custom models. This step involves creating and managing your custom models in Form Recognizer. You can create, train, test, and manage custom models using the Azure portal, REST API, or SDKs.
Step 2: Use the Azure Blob container to upload training data. This step involves uploading your training data to the Azure Blob container. The training data can be in the form of PDFs, images, or other document types that you want to extract data from.
Step 3: Establish your training dataset. This step involves establishing your training dataset by labeling the data. You can use the Form Recognizer labeling tool or other labeling tools to label your training data. Labeling your data involves identifying the regions of interest (ROIs) in the documents and annotating them with the relevant field names.
Step 4: Use labeled or unlabeled datasets to train the model. This step involves training your custom model using your labeled or unlabeled datasets. You can use the Azure portal, REST API, or SDKs to train your model. During training, the model learns to recognize the ROIs and associate them with the field names.
Step 5: Test the model using a dataset not used in training. This step involves testing your model using a dataset that was not used in training. You can use the Azure portal, REST API, or SDKs to test your model. During testing, the model is evaluated for its accuracy, precision, and recall.
Step 6: Create a custom skill and use it in the enrichment pipeline. This step involves creating a custom skill that utilizes your custom model in the enrichment pipeline. You can create custom skills using Azure Cognitive Search or Azure Logic Apps. The custom skill can be used to extract data from documents and enrich the metadata of the documents in the pipeline.
Therefore, the correct sequence of steps is option C: Step 1 -> Step 3 -> Step 2 -> Step 4 -> Step 5 -> Step 6.