Package com.azure.ai.formrecognizer.training

Class FormTrainingClient

java.lang.Object
com.azure.ai.formrecognizer.training.FormTrainingClient
public final class FormTrainingClient extends Object

This class provides a synchronous client to connect to the Form Recognizer Azure Cognitive Service.

This client provides synchronous methods to:

  1. Train a custom model: Train a custom model to analyze and extract data from forms and documents specific to your business using the beginTraining method.
  2. Copy custom model: Copy a custom Form Recognizer model to a target Form Recognizer resource using the beginCopyModel method.
  3. List custom models: Get information about all custom models using the getCustomModel and listCustomModels methods respectively.
  4. Polling and Callbacks: It includes mechanisms for polling the service to check the status of an analysis operation or registering callbacks to receive notifications when the analysis is complete.

Note: This client only supports FormRecognizerServiceVersion.V2_1 and lower. Recommended to use a newer service version, DocumentModelAdministrationAsyncClient and DocumentModelAdministrationClient.

Refer to the Migration guide to use API versions 2022-08-31 and up.

Service clients are the point of interaction for developers to use Azure Form Recognizer. FormTrainingClient is the synchronous service client and FormTrainingAsyncClient is the asynchronous service client. The examples shown in this document use a credential object named DefaultAzureCredential for authentication, which is appropriate for most scenarios, including local development and production environments. Additionally, we recommend using managed identity for authentication in production environments. You can find more information on different ways of authenticating and their corresponding credential types in the Azure Identity documentation".

Sample: Construct a FormTrainingClient with DefaultAzureCredential

The following code sample demonstrates the creation of a FormTrainingClient, using the `DefaultAzureCredentialBuilder` to configure it.

 FormTrainingClient client = new FormTrainingClientBuilder()
     .endpoint("{endpoint}")
     .credential(new DefaultAzureCredentialBuilder().build())
     .buildClient();

Further, see the code sample below to use AzureKeyCredential for client creation.

 FormTrainingClient formTrainingClient = new FormTrainingClientBuilder()
     .credential(new AzureKeyCredential("{key}"))
     .endpoint("{endpoint}")
     .buildClient();
See Also:

  • Method Details

    • getFormRecognizerClient

      public FormRecognizerClient getFormRecognizerClient()
      Creates a new FormRecognizerClient object. The new FormTrainingClient uses the same request policy pipeline as the FormTrainingClient.
      Returns:
      A new FormRecognizerClient object.

    • beginTraining

      public com.azure.core.util.polling.SyncPoller<FormRecognizerOperationResult,CustomFormModel> beginTraining(String trainingFilesUrl, boolean useTrainingLabels)
      Create and train a custom model.

      Models are trained using documents that are of the following content type - 'application/pdf', 'image/jpeg', 'image/png', 'image/tiff'. Other type of content is ignored.

      The service does not support cancellation of the long running operation and returns with an error message indicating absence of cancellation support.

      Code sample

       String trainingFilesUrl = "{SAS-URL-of-your-container-in-blob-storage}";
 boolean useTrainingLabels = true;
 CustomFormModel customFormModel =
     formTrainingClient.beginTraining(trainingFilesUrl, useTrainingLabels).getFinalResult();
 System.out.printf("Model Id: %s%n", customFormModel.getModelId());
 System.out.printf("Model Status: %s%n", customFormModel.getModelStatus());
 customFormModel.getSubmodels()
     .forEach(customFormSubmodel -> customFormSubmodel.getFields()
         .forEach((key, customFormModelField) ->
             System.out.printf("Form Type: %s Field Text: %s Field Accuracy: %f%n",
                 key, customFormModelField.getName(), customFormModelField.getAccuracy())));
      Parameters:
      trainingFilesUrl - an externally accessible Azure storage blob container Uri (preferably a Shared Access Signature Uri). For instructions on setting up forms for training in an Azure Storage Blob Container, see here.
      useTrainingLabels - boolean to specify the use of labeled files for training the model.
      Returns:
      A SyncPoller that polls the training model operation until it has completed, has failed, or has been cancelled. The completed operation returns the trained custom form model.
      Throws:
      FormRecognizerException - If training fails and model with ModelStatus.INVALID is created.
      NullPointerException - If trainingFilesUrl is null.

    • beginTraining

      public com.azure.core.util.polling.SyncPoller<FormRecognizerOperationResult,CustomFormModel> beginTraining(String trainingFilesUrl, boolean useTrainingLabels, TrainingOptions trainingOptions, com.azure.core.util.Context context)
      Create and train a custom model. Models are trained using documents that are of the following content type - 'application/pdf', 'image/jpeg', 'image/png', 'image/tiff'. Other type of content is ignored.

      The service does not support cancellation of the long running operation and returns with an error message indicating absence of cancellation support.

      Code sample

       String trainingFilesUrl = "{SAS-URL-of-your-container-in-blob-storage}";
 TrainingFileFilter trainingFileFilter = new TrainingFileFilter().setSubfoldersIncluded(false).setPrefix("Invoice");
 boolean useTrainingLabels = true;

 CustomFormModel customFormModel = formTrainingClient.beginTraining(trainingFilesUrl, useTrainingLabels,
     new TrainingOptions()
         .setTrainingFileFilter(trainingFileFilter)
         .setPollInterval(Duration.ofSeconds(5)), Context.NONE)
     .getFinalResult();

 System.out.printf("Model Id: %s%n", customFormModel.getModelId());
 System.out.printf("Model Status: %s%n", customFormModel.getModelStatus());
 customFormModel.getSubmodels()
     .forEach(customFormSubmodel -> customFormSubmodel.getFields()
         .forEach((key, customFormModelField) ->
             System.out.printf("Form Type: %s Field Text: %s Field Accuracy: %f%n",
                 key, customFormModelField.getName(), customFormModelField.getAccuracy())));
      Parameters:
      trainingFilesUrl - an externally accessible Azure storage blob container Uri (preferably a Shared Access Signature Uri). For instructions on setting up forms for training in an Azure Storage Blob Container, see here.
      useTrainingLabels - boolean to specify the use of labeled files for training the model.
      trainingOptions - The additional configurable options that may be passed when training a model.
      context - Additional context that is passed through the HTTP pipeline during the service call.
      Returns:
      A SyncPoller that polls the training model operation until it has completed, has failed, or has been cancelled. The completed operation returns the trained custom form model.
      Throws:
      FormRecognizerException - If training fails and model with ModelStatus.INVALID is created.
      NullPointerException - If trainingFilesUrl is null.

    • getCustomModel

      public CustomFormModel getCustomModel(String modelId)
      Get detailed information for a specified custom model id.

      Code sample

       String modelId = "{model_id}";
 CustomFormModel customFormModel = formTrainingClient.getCustomModel(modelId);
 System.out.printf("Model Id: %s%n", customFormModel.getModelId());
 System.out.printf("Model Status: %s%n", customFormModel.getModelStatus());
 customFormModel.getSubmodels()
     .forEach(customFormSubmodel -> customFormSubmodel.getFields()
         .forEach((key, customFormModelField) ->
             System.out.printf("Form Type: %s Field Text: %s Field Accuracy: %f%n",
                 key, customFormModelField.getName(), customFormModelField.getAccuracy())));
      Parameters:
      modelId - The UUID string format model identifier.
      Returns:
      The detailed information for the specified model.
      Throws:
      IllegalArgumentException - If modelId is null or empty.

    • getCustomModelWithResponse

      public com.azure.core.http.rest.Response<CustomFormModel> getCustomModelWithResponse(String modelId, com.azure.core.util.Context context)
      Get detailed information for a specified custom model id.

      Code sample

       String modelId = "{model_id}";
 Response<CustomFormModel> response = formTrainingClient.getCustomModelWithResponse(modelId, Context.NONE);
 System.out.printf("Response Status Code: %d.", response.getStatusCode());
 CustomFormModel customFormModel = response.getValue();
 System.out.printf("Model Id: %s%n", customFormModel.getModelId());
 System.out.printf("Model Status: %s%n", customFormModel.getModelStatus());
 customFormModel.getSubmodels()
     .forEach(customFormSubmodel -> customFormSubmodel.getFields()
         .forEach((key, customFormModelField) ->
             System.out.printf("Field: %s Field Text: %s Field Accuracy: %f%n",
                 key, customFormModelField.getName(), customFormModelField.getAccuracy())));
      Parameters:
      modelId - The UUID string format model identifier.
      context - Additional context that is passed through the Http pipeline during the service call.
      Returns:
      The detailed information for the specified model.
      Throws:
      IllegalArgumentException - If modelId is null or empty.

    • getAccountProperties

      public AccountProperties getAccountProperties()
      Get account information of the form recognizer account.

      Code sample

       AccountProperties accountProperties = formTrainingClient.getAccountProperties();
 System.out.printf("Max number of models that can be trained for this account: %d%n",
     accountProperties.getCustomModelLimit());
 System.out.printf("Current count of trained custom models: %d%n", accountProperties.getCustomModelCount());
      Returns:
      The requested account information of the form recognizer account.

    • getAccountPropertiesWithResponse

      public com.azure.core.http.rest.Response<AccountProperties> getAccountPropertiesWithResponse(com.azure.core.util.Context context)
      Get account information of the form recognizer account with an Http response and a specified Context.

      Code sample

       Response<AccountProperties> response = formTrainingClient.getAccountPropertiesWithResponse(Context.NONE);
 System.out.printf("Response Status Code: %d.", response.getStatusCode());
 AccountProperties accountProperties = response.getValue();
 System.out.printf("Max number of models that can be trained for this account: %s%n",
     accountProperties.getCustomModelLimit());
 System.out.printf("Current count of trained custom models: %d%n", accountProperties.getCustomModelCount());
      Parameters:
      context - Additional context that is passed through the Http pipeline during the service call.
      Returns:
      The requested account information of the form recognizer account.

    • deleteModel

      public void deleteModel(String modelId)
      Deletes the specified custom model.

      Code sample

       String modelId = "{model_id}";
 formTrainingClient.deleteModel(modelId);
 System.out.printf("Model Id: %s is deleted.%n", modelId);
      Parameters:
      modelId - The UUID string format model identifier.
      Throws:
      IllegalArgumentException - If modelId is null or empty.

    • deleteModelWithResponse

      public com.azure.core.http.rest.Response<Void> deleteModelWithResponse(String modelId, com.azure.core.util.Context context)
      Deletes the specified custom model.

      Code sample

       String modelId = "{model_id}";
 Response<Void> response = formTrainingClient.deleteModelWithResponse(modelId, Context.NONE);
 System.out.printf("Response Status Code: %d.", response.getStatusCode());
 System.out.printf("Model Id: %s is deleted.%n", modelId);
      Parameters:
      modelId - The UUID string format model identifier.
      context - Additional context that is passed through the Http pipeline during the service call.
      Returns:
      A Response containing status code and HTTP headers.
      Throws:
      IllegalArgumentException - If modelId is null or empty.

    • listCustomModels

      public com.azure.core.http.rest.PagedIterable<CustomFormModelInfo> listCustomModels()
      List information for each model on the form recognizer account.

      Code sample

       formTrainingClient.listCustomModels()
     .forEach(customModel ->
         System.out.printf("Model Id: %s, Model status: %s, Training started on: %s, Training completed on: %s.%n",
             customModel.getModelId(),
             customModel.getStatus(),
             customModel.getTrainingStartedOn(),
             customModel.getTrainingCompletedOn())
     );
      Returns:
      PagedIterable of CustomFormModelInfo custom form model information.

    • listCustomModels

      public com.azure.core.http.rest.PagedIterable<CustomFormModelInfo> listCustomModels(com.azure.core.util.Context context)
      List information for each model on the form recognizer account with an Http response and a specified Context.

      Code sample

       formTrainingClient.listCustomModels(Context.NONE)
     .forEach(customModel ->
         System.out.printf("Model Id: %s, Model status: %s, Training started on: %s, Training completed on: %s.%n",
             customModel.getModelId(),
             customModel.getStatus(),
             customModel.getTrainingStartedOn(),
             customModel.getTrainingCompletedOn())
     );
      Parameters:
      context - Additional context that is passed through the Http pipeline during the service call.
      Returns:
      PagedIterable of CustomFormModelInfo custom form model information.

    • beginCopyModel

      public com.azure.core.util.polling.SyncPoller<FormRecognizerOperationResult,CustomFormModelInfo> beginCopyModel(String modelId, CopyAuthorization target)
      Copy a custom model stored in this resource (the source) to the user specified target Form Recognizer resource.

      This should be called with the source Form Recognizer resource (with the model that is intended to be copied). The target parameter should be supplied from the target resource's output from FormTrainingAsyncClient.getCopyAuthorization(String, String) method.

      The service does not support cancellation of the long running operation and returns with an error message indicating absence of cancellation support.

      Code sample

       String resourceId = "target-resource-Id";
 String resourceRegion = "target-resource-region";
 String copyModelId = "copy-model-Id";
 formTrainingAsyncClient.getCopyAuthorization(resourceId, resourceRegion)
     .flatMapMany(copyAuthorization -> formTrainingAsyncClient.beginCopyModel(copyModelId, copyAuthorization))
     .flatMap(AsyncPollResponse::getFinalResult)
     .subscribe(customFormModelInfo ->
         System.out.printf("Copied model has model Id: %s, model status: %s, training started on: %s,"
             + " training completed on: %s.%n",
         customFormModelInfo.getModelId(),
         customFormModelInfo.getStatus(),
         customFormModelInfo.getTrainingStartedOn(),
         customFormModelInfo.getTrainingCompletedOn()));
      Parameters:
      modelId - Model identifier of the model to copy to the target Form Recognizer resource
      target - the copy authorization to the target Form Recognizer resource. The copy authorization can be generated from the target resource's call to getCopyAuthorization(String, String)
      Returns:
      A SyncPoller that polls the copy model operation until it has completed, has failed, or has been cancelled.

    • beginCopyModel

      public com.azure.core.util.polling.SyncPoller<FormRecognizerOperationResult,CustomFormModelInfo> beginCopyModel(String modelId, CopyAuthorization target, Duration pollInterval, com.azure.core.util.Context context)
      Copy a custom model stored in this resource (the source) to the user specified target Form Recognizer resource.

      This should be called with the source Form Recognizer resource (with the model that is intended to be copied). The target parameter should be supplied from the target resource's output from getCopyAuthorization(String, String) method.

      The service does not support cancellation of the long running operation and returns with an error message indicating absence of cancellation support.

      Code sample

       // The resource to copy model to
 String resourceId = "target-resource-Id";
 String resourceRegion = "target-resource-region";
 // The Id of the model to be copied
 String copyModelId = "copy-model-Id";

 CopyAuthorization copyAuthorization = targetFormTrainingClient.getCopyAuthorization(resourceId,
     resourceRegion);
 formTrainingClient.beginCopyModel(copyModelId, copyAuthorization, Duration.ofSeconds(5), Context.NONE)
     .waitForCompletion();
 CustomFormModel modelCopy = targetFormTrainingClient.getCustomModel(copyAuthorization.getModelId());
 System.out.printf("Copied model has model Id: %s, model status: %s, was requested on: %s,"
         + " transfer completed on: %s.%n",
     modelCopy.getModelId(),
     modelCopy.getModelStatus(),
     modelCopy.getTrainingStartedOn(),
     modelCopy.getTrainingCompletedOn());
      Parameters:
      modelId - Model identifier of the model to copy to the target Form Recognizer resource
      target - the copy authorization to the target Form Recognizer resource. The copy authorization can be generated from the target resource's call to getCopyAuthorization(String, String)
      pollInterval - Duration between each poll for the operation status. If none is specified, a default of 5 seconds is used.
      context - Additional context that is passed through the HTTP pipeline during the service call.
      Returns:
      A SyncPoller that polls the copy model operation until it has completed, has failed, or has been cancelled.

    • getCopyAuthorization

      public CopyAuthorization getCopyAuthorization(String resourceId, String resourceRegion)
      Generate authorization for copying a custom model into the target Form Recognizer resource.
      Parameters:
      resourceId - Azure Resource Id of the target Form Recognizer resource where the model will be copied to. This information can be found in the 'Properties' section of the Form Recognizer resource in the Azure Portal.
      resourceRegion - Location of the target Form Recognizer resource. A valid Azure region name supported by Cognitive Services. This information can be found in the 'Keys and Endpoint' section of the Form Recognizer resource in the Azure Portal.

      Code sample

       String resourceId = "target-resource-Id";
 String resourceRegion = "target-resource-region";
 CopyAuthorization copyAuthorization = formTrainingClient.getCopyAuthorization(resourceId, resourceRegion);
 System.out.printf("Copy Authorization for model id: %s, access token: %s, expiration time: %s, "
         + "target resource Id; %s, target resource region: %s%n",
     copyAuthorization.getModelId(),
     copyAuthorization.getAccessToken(),
     copyAuthorization.getExpiresOn(),
     copyAuthorization.getResourceId(),
     copyAuthorization.getResourceRegion()
 );
      Returns:
      The CopyAuthorization

    • getCopyAuthorizationWithResponse

      public com.azure.core.http.rest.Response<CopyAuthorization> getCopyAuthorizationWithResponse(String resourceId, String resourceRegion, com.azure.core.util.Context context)
      Generate authorization for copying a custom model into the target Form Recognizer resource. This should be called by the target resource (where the model will be copied to) and the output can be passed as the target parameter into FormTrainingAsyncClient.beginCopyModel(String, CopyAuthorization).
      Parameters:
      resourceId - Azure Resource Id of the target Form Recognizer resource where the model will be copied to. This information can be found in the 'Properties' section of the Form Recognizer resource in the Azure Portal.
      resourceRegion - Location of the target Form Recognizer resource. A valid Azure region name supported by Cognitive Services.This information can be found in the 'Keys and Endpoint' section of the Form Recognizer resource in the Azure Portal.
      context - Additional context that is passed through the Http pipeline during the service call.

      Code sample

       String resourceId = "target-resource-Id";
 String resourceRegion = "target-resource-region";
 Response<CopyAuthorization> copyAuthorizationResponse =
     formTrainingClient.getCopyAuthorizationWithResponse(resourceId, resourceRegion, Context.NONE);
 System.out.printf("Copy Authorization operation returned with status: %s",
     copyAuthorizationResponse.getStatusCode());
 CopyAuthorization copyAuthorization = copyAuthorizationResponse.getValue();
 System.out.printf("Copy model id: %s, access token: %s, expiration time: %s, "
         + "target resource Id; %s, target resource region: %s%n",
     copyAuthorization.getModelId(),
     copyAuthorization.getAccessToken(),
     copyAuthorization.getExpiresOn(),
     copyAuthorization.getResourceId(),
     copyAuthorization.getResourceRegion()
 );
      Returns:
      A Response containing the CopyAuthorization

    • beginCreateComposedModel

      public com.azure.core.util.polling.SyncPoller<FormRecognizerOperationResult,CustomFormModel> beginCreateComposedModel(List<String> modelIds)
      Create a composed model from the provided list of existing models in the account.

      This operations fails if the list consists of an invalid, non-existing model Id or duplicate Ids. This operation is currently only supported for custom models trained using labels.

      The service does not support cancellation of the long running operation and returns with an error message indicating absence of cancellation support.

      Code sample

       String labeledModelId1 = "5f21ab8d-71a6-42d8-9856-ef5985c486a8";
 String labeledModelId2 = "d7b0904c-841f-46f9-a9f4-3f2273eef7c9";
 final CustomFormModel customFormModel
     = formTrainingClient.beginCreateComposedModel(Arrays.asList(labeledModelId1, labeledModelId2))
     .getFinalResult();
 System.out.printf("Model Id: %s%n", customFormModel.getModelId());
 System.out.printf("Model Status: %s%n", customFormModel.getModelStatus());
 System.out.printf("Is this a composed model: %s%n",
     customFormModel.getCustomModelProperties().isComposed());
 customFormModel.getSubmodels()
     .forEach(customFormSubmodel -> customFormSubmodel.getFields()
         .forEach((key, customFormModelField) ->
             System.out.printf("Form type: %s Field Text: %s Field Accuracy: %f%n",
                 key, customFormModelField.getName(), customFormModelField.getAccuracy())));
      Parameters:
      modelIds - The list of models Ids to form the composed model.
      Returns:
      A SyncPoller that polls the create composed model operation until it has completed, has failed, or has been cancelled. The completed operation returns the composed model.
      Throws:
      FormRecognizerException - If create composed model operation fails and model with OperationStatus.FAILED is created.
      NullPointerException - If the list of modelIds is null or empty.

    • beginCreateComposedModel

      public com.azure.core.util.polling.SyncPoller<FormRecognizerOperationResult,CustomFormModel> beginCreateComposedModel(List<String> modelIds, CreateComposedModelOptions createComposedModelOptions, com.azure.core.util.Context context)
      Create a composed model from the provided list of existing models in the account.

      This operations fails if the list consists of an invalid, non-existing model Id or duplicate Ids. This operation is currently only supported for custom models trained using labels.

      The service does not support cancellation of the long running operation and returns with an error message indicating absence of cancellation support.

      Code sample

       String labeledModelId1 = "5f21ab8d-71a6-42d8-9856-ef5985c486a8";
 String labeledModelId2 = "d7b0904c-841f-46f9-a9f4-3f2273eef7c9";
 final CustomFormModel customFormModel =
     formTrainingClient.beginCreateComposedModel(Arrays.asList(labeledModelId1, labeledModelId2),
         new CreateComposedModelOptions()
             .setModelName("my composed model name"),
         Context.NONE)
         .setPollInterval(Duration.ofSeconds(5))
         .getFinalResult();
 System.out.printf("Model Id: %s%n", customFormModel.getModelId());
 System.out.printf("Model Status: %s%n", customFormModel.getModelStatus());
 System.out.printf("Model display name: %s%n", customFormModel.getModelName());
 System.out.printf("Is this a composed model: %s%n",
     customFormModel.getCustomModelProperties().isComposed());
 customFormModel.getSubmodels()
     .forEach(customFormSubmodel -> customFormSubmodel.getFields()
         .forEach((key, customFormModelField) ->
             System.out.printf("Form type: %s Field Text: %s Field Accuracy: %f%n",
                 key, customFormModelField.getName(), customFormModelField.getAccuracy())));
      Parameters:
      modelIds - The list of models Ids to form the composed model.
      createComposedModelOptions - The configurable options to pass when creating a composed model.
      context - Additional context that is passed through the HTTP pipeline during the service call.
      Returns:
      A SyncPoller that polls the create composed model operation until it has completed, has failed, or has been cancelled. The completed operation returns the composed model.
      Throws:
      FormRecognizerException - If create composed model operation fails and model with OperationStatus.FAILED is created.
      NullPointerException - If the list of modelIds is null or empty.