How to Integrate Microsoft Azure Cognitive Services Custom Vision Prediction API in C#

Integrate Microsoft Azure Cognitive Services Custom Vision Prediction API in C#

• Create a new project in Visual Studio. For this example, choose a Console App (.NET Core) to clearly understand the integration without any UI complexities.

 

• Install the necessary NuGet packages required for making HTTP requests. Use the HttpClient class for this purpose. You can install these using the Package Manager Console:

Install-Package Microsoft.Extensions.Configuration
Install-Package Microsoft.Extensions.Configuration.Json

• Include namespaces in your Program.cs file. At the beginning of the file, ensure you have the proper namespaces:

using System;
using System.IO;
using System.Net.Http;
using System.Net.Http.Headers;
using System.Threading.Tasks;
using Microsoft.Extensions.Configuration;

• Load Configuration to handle sensitive information like API endpoints and keys externally:

//appsettings.json
{
  "PredictionEndpoint": "your_prediction_endpoint",
  "PredictionKey": "your_prediction_key",
  "ProjectId": "your_project_id",
  "PublishedName": "your_published_name"
}

//In your Program.cs
IConfiguration config = new ConfigurationBuilder()
    .SetBasePath(Directory.GetCurrentDirectory())
    .AddJsonFile("appsettings.json", optional: true, reloadOnChange: true)
    .Build();

• Create a method that communicates with the Custom Vision Prediction API:

static async Task PredictImageUrlAsync(IConfiguration config, string imageUrl)
{
    var client = new HttpClient();
    var endpoint = config["PredictionEndpoint"];
    var predictionKey = config["PredictionKey"];
    var projectId = config["ProjectId"];
    var publishedName = config["PublishedName"];

    client.DefaultRequestHeaders.Add("Prediction-Key", predictionKey);

    string url = $"{endpoint}/{projectId}/url?iterationId={publishedName}";

    var content = new StringContent($"{{\"Url\": \"{imageUrl}\"}}", System.Text.Encoding.UTF8, "application/json");
    HttpResponseMessage response = await client.PostAsync(url, content);

    if (response.IsSuccessStatusCode)
    {
        Console.WriteLine(await response.Content.ReadAsStringAsync());
    }
    else
    {
        Console.WriteLine($"Error: {response.StatusCode}");
    }
}

• Implement the Main method to invoke image prediction:

async Task Main(string[] args)
{
    var config = new ConfigurationBuilder()
        .SetBasePath(Directory.GetCurrentDirectory())
        .AddJsonFile("appsettings.json", optional: true, reloadOnChange: true)
        .Build();

    string imageUrl = "https://example.com/image.jpg"; // Replace with the actual image URL
    await PredictImageUrlAsync(config, imageUrl);
}

• Run the application to predict the given image URL using Azure Custom Vision Prediction API.

Handling Responses

• To properly handle responses, consider deserializing the JSON response into C# objects. Use libraries such as Newtonsoft.Json:

Install-Package Newtonsoft.Json

// Add using directives
using Newtonsoft.Json;

• Create a class structure to match the JSON response structure and use JsonConvert.DeserializeObject to parse it:

public class Prediction
{
    public string TagName { get; set; }
    public double Probability { get; set; }
}

public class PredictionResult
{
    public List<Prediction> Predictions { get; set; }
}

static async Task ProcessResponseAsync(HttpResponseMessage response)
{
    if(response.IsSuccessStatusCode)
    {
        var responseData = await response.Content.ReadAsStringAsync();
        var result = JsonConvert.DeserializeObject<PredictionResult>(responseData);
        foreach(var prediction in result.Predictions)
        {
            Console.WriteLine($"{prediction.TagName}: {prediction.Probability:P2}");
        }
    }
    else
    {
        Console.WriteLine($"Error: {response.StatusCode}");
    }
}