Azure AI services - Document intelligence - Extract Data from Forms

 Azure AI services - Document intelligence - Extract Data from Forms:

Source:

https://microsoftlearning.github.io/AI-102-AIEngineer/

https://github.com/MicrosoftLearning/AI-102-AIEngineer

1. Create Azure AI services - Document intelligence in Azure Portal
2. Run Powershell script to create Storage account
3. Train the Model
4. Test the Model 

Required Dlls;

dotnet add package Azure.Core --version 1.44.1

dotnet add package Azure.AI.FormRecognizer --version 4.1.0

dotnet add package Azure.AI.FormRecognizer --version 3.0.0

dotnet add package Tabulate.NET --version 1.0.5

2. Run Powershell script to create Storage account

@echo off

SETLOCAL ENABLEDELAYEDEXPANSION

rem Set variable values

set subscription_id=129b2bb6-asdf-asdf-83ba-85bf570bebca

set resource_group=rg1

set location=eastus

set expiry_date=2026-01-01T00:00:00Z

rem Get random numbers to create unique resource names

set unique_id=!random!!random!

rem Create a storage account in your Azure resource group

echo Creating storage...

call az storage account create --name ai102form!unique_id! --subscription !subscription_id! --resource-group !resource_group! --location !location! --sku Standard_LRS --encryption-services blob --default-action Allow --output none --allow-blob-public-access true

echo Uploading files...

rem Get storage key to create a container in the storage account

for /f "tokens=*" %%a in (

'az storage account keys list --subscription !subscription_id! --resource-group !resource_group! --account-name ai102form!unique_id! --query "[?keyName=='key1'].{keyName:keyName, permissions:permissions, value:value}"'

) do (

set key_json=!key_json!%%a

)

set key_string=!key_json:[ { "keyName": "key1", "permissions": "Full", "value": "=!

set AZURE_STORAGE_KEY=!key_string:" } ]=!

rem Create container

call az storage container create --account-name ai102form!unique_id! --name sampleforms --auth-mode key --account-key %AZURE_STORAGE_KEY% --output none

rem Upload files from your local sampleforms folder to a container called sampleforms in the storage account

rem Each file is uploaded as a blob

call az storage blob upload-batch -d sampleforms -s ./sample-forms --account-name ai102form!unique_id! --auth-mode key --account-key %AZURE_STORAGE_KEY%  --output none

rem Set a variable value for future use

set STORAGE_ACCT_NAME=ai102form!unique_id!

rem Get a Shared Access Signature (a signed URI that points to one or more storage resources) for the blobs in sampleforms  

for /f "tokens=*" %%a in (

'az storage container generate-sas --account-name ai102form!unique_id! --name sampleforms --expiry !expiry_date! --permissions rwl'

) do (

set SAS_TOKEN=%%a

set SAS_TOKEN=!SAS_TOKEN:~1,-1!

)

set URI=https://!STORAGE_ACCT_NAME!.blob.core.windows.net/sampleforms?!SAS_TOKEN!

rem Print the generated Shared Access Signature URI, which is used by Azure Storage to authorize access to the storage resource

echo -------------------------------------

echo SAS URI: !URI!

Run the code : dotnet run

OutPut:

3. Train the Model

using System;

using System.IO;

using System.Collections.Generic;

using System.Threading.Tasks;

using Microsoft.Extensions.Configuration;

// import namespaces

using Azure;

using Azure.AI.FormRecognizer;

using Azure.AI.FormRecognizer.Models;

using Azure.AI.FormRecognizer.Training;

namespace train_model

{

    class Program

    {

        static async Task Main(string[] args)

        {

            try

            {

                // Get configuration settings

                // IConfigurationBuilder builder = new ConfigurationBuilder().AddJsonFile("appsettings.json");

                // IConfigurationRoot configuration = builder.Build();

                // string formEndpoint = configuration["FormEndpoint"];

                // string formKey = configuration["FormKey"];

                // string trainingStorageUri = configuration["StorageUri"];

                // "YOUR_FORM_RECOGNIZER_ENDPOINT"

                string formEndpoint = "https://a.cognitiveservices.azure.com/";

                // "YOUR_FORM_RECOGNIZER_KEY"

                string formKey = "1E7gEDsZ2pUAiximoBAACYeBjFXJ3w3AAALACOGu5mm";

                // "YOUR_SAS_URI"

                string trainingStorageUri = "https://8IqQY2WOeCJRHTPFg%3D";

                // Authenticate Form Training Client

                var credential = new AzureKeyCredential(formKey);

                var trainingClient = new FormTrainingClient(new Uri(formEndpoint), credential);

                // Train model

                CustomFormModel model = await trainingClient

                .StartTrainingAsync(new Uri(trainingStorageUri), useTrainingLabels: true)

                .WaitForCompletionAsync();

                // Get model info

                Console.WriteLine($"Custom Model Info:");

                Console.WriteLine($"    Model Id: {model.ModelId}");

                Console.WriteLine($"    Model Status: {model.Status}");

                Console.WriteLine($"    Training model started on: {model.TrainingStartedOn}");

                Console.WriteLine($"    Training model completed on: {model.TrainingCompletedOn}");

            }

            catch (Exception ex)

            {

                Console.WriteLine(ex.Message);

            }

        }

    }

}

Run the code : dotnet run

OutPut:

4. Test the Model 

using System;

using System.IO;

using System.Collections.Generic;

using System.Threading.Tasks;

using Microsoft.Extensions.Configuration;

// import namespaces

using Azure;

using Azure.AI.FormRecognizer;

using Azure.AI.FormRecognizer.Models;

using Azure.AI.FormRecognizer.Training;

namespace test_model

{

    class Program

    {

        static async Task Main(string[] args)

        {

            try

            {

                // Get configuration settings from AppSettings

                // IConfigurationBuilder builder = new ConfigurationBuilder().AddJsonFile("appsettings.json");

                // IConfigurationRoot configuration = builder.Build();

                // string formEndpoint = configuration["FormEndpoint"];

                // string formKey = configuration["FormKey"];

                // string modelId = configuration["ModelId"];

                // "YOUR_FORM_RECOGNIZER_ENDPOINT";                 string formEndpoint = "https://a.cognitiveservices.azure.com/";

                // "YOUR_FORM_RECOGNIZER_KEY";                 string formKey = "1E7gEDsZ2pUAiximAAALACOGu5mm";

                // "YOUR_MODEL_ID";

                string modelId = "7891e019-9cc2-48a9-a9e6-08ac408484c5";

                // Authenticate Azure AI Document Intelligence Client

                var credential = new AzureKeyCredential(formKey);

                var recognizerClient = new FormRecognizerClient(new Uri(formEndpoint), credential);

                // Get form url for testing  

                string image_file = "test1.jpg";

                using (var image_data = File.OpenRead(image_file))

                {

                    // Use trained model with new form

                    RecognizedFormCollection forms = await recognizerClient

                    .StartRecognizeCustomForms(modelId, image_data)

                    .WaitForCompletionAsync();

                    foreach (RecognizedForm form in forms)

                    {

                        Console.WriteLine($"Form of type: {form.FormType}");

                        foreach (FormField field in form.Fields.Values)

                        {

                            Console.WriteLine($"Field '{field.Name}':");

                            if (field.LabelData != null)

                            {

                                Console.WriteLine($"    Label: '{field.LabelData.Text}'");

                            }

                            Console.WriteLine($"    Value: '{field.ValueData.Text}'");

                            Console.WriteLine($"    Confidence: {field.Confidence}");

                        }

                    }

                }

            }

            catch (Exception ex)

            {

                Console.WriteLine(ex.Message);

            }

        }

    }

}

OutPut: