UiPath Integrations

Overview

Each Azure Form Recognizer activity calls a Form Recognizer API. To successfully call the API, the activities must include an authentication header in the request. This header includes a subscription key that validates your subscription to the Azure Form Recognizer service.

To generate a subscription key and start building your automation projects, complete the following:

  1. Create a Form Recognizer Azure resource
  2. Build and upload a training data set
  3. Get shared access signature
  4. Build your project

1. Create a Form Recognizer Azure resource

Complete Microsoft's documented steps to create a Form Recognizer resource.

Note

The Azure Form Recognizer service is available in a limited-access preview. Because of this, the process to create a resource may be different when compared to other Azure services (to learn more, see request access in the Microsoft documentation).

When your Form Recognizer resource finishes deployment:

  1. Find and select it from the All resources list in the portal.
  2. Click the Quick start tab under RESOURCE MANAGEMENT to view your subscription data.
  3. Note the values of Key1 and Endpoint (you'll enter these in the Azure Form Recognizer Scope activity).

2. Build and upload a training data set

Follow Microsoft's documented tips and input requirements to build a training data set for a custom model.

Note

To build a training data set for our examples, we used the files in the Train folder of Microsoft's provided sample data set.

After building your training data set, follow Microsoft's documented steps to create a container (you may need to create a storage account first if you don't have one).

After creating your container:

  1. Go to Storage Explorer in the Azure portal.
  2. Under Subscriptions, find and select the storage account that you created your container in.
  3. Under the account's Blob containers, select the container that you created.
  4. Upload your training data set (e.g., 5 PDF invoices).

3. Get shared access signature

The Train Model activity requires a URI to the container that has your training data set.

To get the container URI:

  1. Go to Storage Explorer in the Azure portal.
  2. Under Subscriptions, find and select the storage account that you created your container in.
  3. Under the account's Blob containers, right-click the container that you created.
  4. Click Get Shared Access Signature (this opens Shared Access Signature window).
  5. Enter your Access policy, Start time, Expiry time, Time zone, and Permissions.
    • At a minimum, select Read and List from the Permissions list.
  6. Click Create.
  7. Note the URI value.

4. Build your project

  1. Create a new automation project.
    1. Open UiPath Studio.
    2. Under New Project, click Process (this opens a New Blank Process window).
    3. Enter a project Name, Location, and Description.
    4. Click Create.
  1. Install the UiPath.AzureFormRecognizer.Activities package.
    1. In the Design ribbon, click Manage Packages (this opens the Manage Packages window).
    2. Under All Packages, click Go.
    3. In the Search bar, enter Azure.
    4. Click, install and accept the license for the UiPath.AzureFormRecognizer.Activities.
      • For more information about Uipath Studio packages, see Managing Packages in the Studio Guide.

You're done!

Now that you have completed the setup, you can start adding the Azure Form Recognizer activities to your project.

Next steps

For a hands-on learning experience and to quickly start using the activities, see the Quickstart guide. This guide provides step-by-step instructions to help you create a working sample of the different activities so that you can verify the connection to your Form Recognizer Azure resource and get familiar with the input/output properties.

To learn more about the Azure Form Recognizer activities (including example property inputs/outputs), see the Activities page for a complete list of activities and links to the detail pages.

Updated 5 days ago


Setup


Suggested Edits are limited on API Reference Pages

You can only suggest edits to Markdown body content, but not to the API spec.