This project demonstrates how to create a process that automatically extracts information from forms using Azure Form Recognizer and Azure Logic Apps.
- When a file is uploaded to a Sharepoint folder, an Azure Logic App is triggered.
- The Logic App will call Form Recognizer to analyze the form (in this case, using a custom model of Azure Form Recognizer). The same can be done with the default Layout or Document APIs from Form Recognizer.
- The Logic App will store the results of the Form Recognizer in Blob storage, where they can be used by another application or process.
NOTE: The Azure Logic App Form Recognizer connector is compatible with Form Recognizer V2.0 (custom models built with the FOTT tool), not with the newer Form Recognizer V3.0 resource. Ensure you are training your custom models with the FOTT tool.
-
Create Sharepoint Site
-
Deploy Azure Resources
Option 1: (note that you may have to configure the Form Recognizer resource according to the steps documented in the Microsoft Docs to get access to the FOTT tool to create a custom model
Option 2: Step-by-step process
- Create a Logic App
- Create a Form Recognizer resource. Note that we will opt for v2.1 of the Form Recognizer as it is the latest General Availability and the Logic App connector we will use has been built for v2.1. If you plan on using a custom model (as I am doing here), follow the steps documented in the Microsoft Docs
- Create a Storage Account. Within the Storage Account, create 2 containers, one for training documents to train the Form Recognizer Custom Model (ex: trainingdocuments) and another container for the export from Logic App (ex: exportedresults)
-
(If you are planning to use a Form Recognizer custom model) Train your custom model using at least 5 documents using the FOTT tool
-
Configure your Logic App
- Select the Sharepoint connector
When a file is created in a folder
. - Add a Form Recognizer action
Analyze Custom Form
. - Add a Blob Storage action
Create blob (V2)
.
- Select the Sharepoint connector