How to Use Microsoft Azure Form Recognizer API in .NET

Getting Started with Azure Form Recognizer API

• Before diving into the code, ensure that you have set up your Azure Form Recognizer resources, and obtained the endpoint and API key from the Azure portal.

 

• Make sure you have installed the Azure.AI.FormRecognizer nuget package in your .NET project to help interact with the API.

 

• You'll typically work with the FormRecognizerClient class, which provides methods to analyze documents.

dotnet add package Azure.AI.FormRecognizer

Sample Configuration and Initialization

• Initialize the FormRecognizerClient using the endpoint and API key obtained from your Azure account.

 

• You'll usually initialize this client in a part of your application responsible for handling dependency injection or service initialization.

using Azure;
using Azure.AI.FormRecognizer;
using Azure.AI.FormRecognizer.Models;
using Azure.AI.FormRecognizer.DocumentAnalysis;
using System;
using System.IO;
using System.Threading.Tasks;

namespace FormRecognizerExample
{
    public class FormRecognizerService
    {
        private readonly FormRecognizerClient _formRecognizerClient;

        public FormRecognizerService(string endpoint, string apiKey)
        {
            var credential = new AzureKeyCredential(apiKey);
            _formRecognizerClient = new FormRecognizerClient(new Uri(endpoint), credential);
        }
    }
}

Analyzing Forms with Form Recognizer

• To analyze forms, you'll often use the StartRecognizeContentAsync method for basic text extraction or the StartRecognizeCustomFormsAsync for more complex form processing.

 

• For custom form models, make sure you have a form model ID from the forms you have trained previously in Azure.

public async Task AnalyzeFormAsync(string filePath)
{
    using var stream = new FileStream(filePath, FileMode.Open);

    var operation = await _formRecognizerClient.StartRecognizeContentAsync(stream);
    var result = await operation.WaitForCompletionAsync();

    foreach (var page in result.Value)
    {
        Console.WriteLine($"Page number: {page.PageNumber}");

        foreach (var table in page.Tables)
        {
            Console.WriteLine("Table data:");

            foreach (var cell in table.Cells)
            {
                Console.WriteLine($"Cell text: '{cell.Content}'");
            }
        }
    }
}

Handling Models and Training Data

• Form Recognizer allows training models on your custom forms. This involves uploading training data and labeling attributes that the models will learn to recognize.

 

• Make sure your training data is clean and well-organized to improve the accuracy of the form recognizer model.

public async Task TrainModelAsync(string trainingDataUrl)
{
    var options = new BuildModelOptions { };
    var operation = await _formRecognizerClient.StartBuildModelAsync(trainingDataUrl, options);
    var customFormModel = await operation.WaitForCompletionAsync();

    Console.WriteLine($"Model ID: {customFormModel.Value.ModelId}");
}

Handling Errors and Debugging

• When working with live APIs, error handling is crucial. Use try-catch blocks and inspect exceptions for better reliability and debugging.

 

• Azure's SDK provides extensive logging capabilities which can help debug and identify issues in the interaction with the API.

try
{
    await AnalyzeFormAsync("path/to/your/document.pdf");
}
catch (Exception ex)
{
    Console.WriteLine($"An error occurred: {ex.Message}");
}

Optimizing Performance

• Consider optimizing network calls by batching requests or using asynchronous operations, as shown in the examples, to maximize application responsiveness.

 

• Keep models updated with new data samples to improve accuracy and account for data drift over time.