AWS vs Google Vision（OCR 功能比较）

2.1. Key Features of AWS Textract

• Text Extraction: Textract accurately extracts text from diverse document types, such as scanned paper documents, forms, and invoices.
• Form and Table Extraction: It identifies and extracts structured data from forms and tables, preserving the original layout and formatting.
• Integration with Other AWS Services: Textract seamlessly integrates with various AWS services, facilitating automated workflows and enhanced data processing.

2.2. Licensing

AWS Textract operates on a pay-as-you-go pricing model, where users are billed based on the number of pages processed.

2.3. Installation

Before utilizing Amazon Textract for the first time, follow these steps:

1. Register for AWS Services:

   • Sign up for an AWS account to access Amazon Textract and related services.
2. Establish an IAM User:
   • Create an IAM (Identity and Access Management) user with appropriate permissions for accessing Amazon Textract.

Once you've completed the account setup and IAM user creation, proceed to configure access keys within the AWS console to programmatically access the API using C#. You'll need the following:

• AccessKeyId
• SecretAccessKey
• RegionEndPoint (Your access area)

In this example, the endpoint PKISB1 is used.

Now create a new Visual Studio Project. Then go to the Tools menu and select the NuGet Package Manager and choose Manage NuGet Packages for Solutions.

In the search box enter "AWSSDK" and install the latest version.

2.4. Code Example (Using AWS SDK for .NET)

// Import necessary AWS SDK namespaces
using Amazon;
using Amazon.Textract;
using Amazon.Textract.Model;

// Create a new Textract client using your AWS credentials and region
var client = new AmazonTextractClient("your_access_key_id", "your_secret_access_key", Amazon.RegionEndpoint.PKISB1);

// Prepare a request to analyze a document in an S3 bucket
var request = new AnalyzeDocumentRequest
{
    Document = new Document
    {
        S3Object = new S3Object
        {
            Bucket = "your-bucket-name",
            Name = "your-document-key"
        }
    },
    FeatureTypes = new List<string> { "FORMS", "TABLES" }
};

// Call the AnalyzeDocumentAsync method to asynchronously analyze the document
var response = await client.AnalyzeDocumentAsync(request);

// Import necessary AWS SDK namespaces
using Amazon;
using Amazon.Textract;
using Amazon.Textract.Model;

// Create a new Textract client using your AWS credentials and region
var client = new AmazonTextractClient("your_access_key_id", "your_secret_access_key", Amazon.RegionEndpoint.PKISB1);

// Prepare a request to analyze a document in an S3 bucket
var request = new AnalyzeDocumentRequest
{
    Document = new Document
    {
        S3Object = new S3Object
        {
            Bucket = "your-bucket-name",
            Name = "your-document-key"
        }
    },
    FeatureTypes = new List<string> { "FORMS", "TABLES" }
};

// Call the AnalyzeDocumentAsync method to asynchronously analyze the document
var response = await client.AnalyzeDocumentAsync(request);

' Import necessary AWS SDK namespaces
Imports Amazon
Imports Amazon.Textract
Imports Amazon.Textract.Model

' Create a new Textract client using your AWS credentials and region
Private client = New AmazonTextractClient("your_access_key_id", "your_secret_access_key", Amazon.RegionEndpoint.PKISB1)

' Prepare a request to analyze a document in an S3 bucket
Private request = New AnalyzeDocumentRequest With {
	.Document = New Document With {
		.S3Object = New S3Object With {
			.Bucket = "your-bucket-name",
			.Name = "your-document-key"
		}
	},
	.FeatureTypes = New List(Of String) From {"FORMS", "TABLES"}
}

' Call the AnalyzeDocumentAsync method to asynchronously analyze the document
Private response = await client.AnalyzeDocumentAsync(request)

3.1. Key Features of Google Vision

• OCR and Text Detection: Google Vision accurately detects and extracts text from images and documents, supporting multiple languages.
• Image Analysis: It offers various image analysis capabilities, including label detection, face detection, and landmark detection.
• Integration with Google Cloud Services: Google Vision can be seamlessly integrated with other Google Cloud services to create comprehensive solutions.

3.2. Licensing

Google Vision operates on a pay-as-you-go pricing model, and users are billed based on the number of units (e.g., data entry images, text, etc.) processed.

3.3. Installation

To integrate the Vision API into your C# project, ensure you complete these necessary steps:

1. Establish a Google Account.
2. Generate a new project via the Google Cloud Console.
3. Activate billing for the project.
4. Enable the Vision API.
5. Generate a Service Account and configure the associated credentials.
6. Download the service account key credentials in JSON file format.

Once the credentials are downloaded, create a new project in Visual Studio and install the Google Cloud Platform (Google Vision) SDK using the NuGet Package Manager.

3.4. Code Example (Using Google Cloud Client Libraries)

// Import necessary Google Cloud Vision namespaces
using Google.Cloud.Vision.V1;
using Google.Protobuf;
using System.IO;
using Google.Apis.Auth.OAuth2;

// Load the service account credentials from the JSON file
var credential = GoogleCredential.FromFile("path-to-credentials.json");
var clientBuilder = new ImageAnnotatorClientBuilder { CredentialsPath = "path-to-credentials.json" };

// Build the ImageAnnotatorClient using the credentials
var client = clientBuilder.Build();

// Load an image file for text detection
var image = Image.FromFile("path-to-your-image.jpg");

// Perform text detection on the image
var response = client.DetectText(image);

// Output the detected text descriptions
foreach (var annotation in response)
{
    Console.WriteLine(annotation.Description);
}

// Import necessary Google Cloud Vision namespaces
using Google.Cloud.Vision.V1;
using Google.Protobuf;
using System.IO;
using Google.Apis.Auth.OAuth2;

// Load the service account credentials from the JSON file
var credential = GoogleCredential.FromFile("path-to-credentials.json");
var clientBuilder = new ImageAnnotatorClientBuilder { CredentialsPath = "path-to-credentials.json" };

// Build the ImageAnnotatorClient using the credentials
var client = clientBuilder.Build();

// Load an image file for text detection
var image = Image.FromFile("path-to-your-image.jpg");

// Perform text detection on the image
var response = client.DetectText(image);

// Output the detected text descriptions
foreach (var annotation in response)
{
    Console.WriteLine(annotation.Description);
}

' Import necessary Google Cloud Vision namespaces
Imports Google.Cloud.Vision.V1
Imports Google.Protobuf
Imports System.IO
Imports Google.Apis.Auth.OAuth2

' Load the service account credentials from the JSON file
Private credential = GoogleCredential.FromFile("path-to-credentials.json")
Private clientBuilder = New ImageAnnotatorClientBuilder With {.CredentialsPath = "path-to-credentials.json"}

' Build the ImageAnnotatorClient using the credentials
Private client = clientBuilder.Build()

' Load an image file for text detection
Private image = System.Drawing.Image.FromFile("path-to-your-image.jpg")

' Perform text detection on the image
Private response = client.DetectText(image)

' Output the detected text descriptions
For Each annotation In response
	Console.WriteLine(annotation.Description)
Next annotation

4.1. Key Features of IronOCR

• On-Premises OCR: IronOCR enables on-premises text extraction by integrating OCR functionality into applications.
• Versatile Language Support: It supports a wide range of languages (125+ International Languages).
• Advanced Text Recognition: IronOCR offers advanced text recognition capabilities, including font and style detection, and handles various image formats.

4.2. Licensing

IronOCR offers a full server framework and a variety of licensing options, including a free trial and paid licenses based on your application server usage and deployment needs.

4.3. Installation

Installing IronOCR is a straightforward process. Create a new Visual Studio Project and open the NuGet Package Manager for Solutions, search "IronOCR". A list will appear; select the latest version of IronOCR and click on Install.

4.4. Code Example (C#)

// Import the IronOcr namespace
using IronOcr;

// Initialize the IronTesseract OCR engine
var ocr = new IronTesseract();
ocr.Language = OcrLanguage.English;

// Read and extract text from an image file
var result = ocr.Read("path-to-your-image.jpg");

// Output the extracted text
Console.WriteLine(result.Text);

// Import the IronOcr namespace
using IronOcr;

// Initialize the IronTesseract OCR engine
var ocr = new IronTesseract();
ocr.Language = OcrLanguage.English;

// Read and extract text from an image file
var result = ocr.Read("path-to-your-image.jpg");

// Output the extracted text
Console.WriteLine(result.Text);

' Import the IronOcr namespace
Imports IronOcr

' Initialize the IronTesseract OCR engine
Private ocr = New IronTesseract()
ocr.Language = OcrLanguage.English

' Read and extract text from an image file
Dim result = ocr.Read("path-to-your-image.jpg")

' Output the extracted text
Console.WriteLine(result.Text)

a. Precision and Efficiency

• AWS Textract and Google Vision, being cloud-centric solutions, harness potent machine learning models and boast commendable precision in text extraction.
• IronOCR, a potent software library, stands out as a winner in terms of precision and efficiency, provided it's effectively integrated into the application.

b. User-Friendliness and Seamless Integration

• AWS Textract and Google Vision offer easy integration via APIs, ensuring a streamlined process for developers.
• However, IronOCR, while exceptionally versatile, necessitates integration into the application's codebase, demanding a bit more custom development effort.

c. Scalability

• AWS Textract and Google Vision exhibit exceptional scalability as cloud services, effortlessly managing substantial request volumes.
• In comparison, IronOCR's scalability is contingent upon the application's infrastructure and its ability to handle OCR processing within the application itself.

d. Financial Considerations

• AWS Textract and Google Vision follow a pay-as-you-go pricing model, potentially rendering them cost-effective based on usage.
• Contrastingly, IronOCR typically involves a one-time purchase or subscription-based model, presenting long-term cost-efficiency benefits, making it a standout winner.