I have a program in .Net Core to describe an image using Azure OpenAPI GPT-4o model.

using Azure;
using Azure.AI.OpenAI;
using OpenAI;
using OpenAI.Chat;
using System;
using System.ClientModel;
using System.Collections.Generic;
using System.IO;
using System.Threading.Tasks;

// Azure OpenAI client library for .NET - version 2.1.0

public class AzureOpenAiService : IAzureOpenAiService
{
    private static string endpoint = "/";
    private static string deployment = "gpt-4o";
    private static string apiKey = "LFK";

    public async Task FindPrimarySubjectAsync(string imagePath)
    {
        try
        {
            string base64Image = EncodeImage(imagePath);

            var credential = new AzureKeyCredential(apiKey);
            OpenAIClientOptions openAIClientOptions = new OpenAIClientOptions
            {
                Endpoint = new Uri(endpoint)
            };

            var client = new AzureOpenAIClient(new Uri(endpoint), new ApiKeyCredential(apiKey));

            var chatMessages = new List<ChatMessage>
            {
                new SystemChatMessage("Analyze the uploaded image and return a single-word description of the main subject. The response should be only one word, representing the most general yet accurate category."),
                new UserChatMessage($"What is in this image? Image: data:image/png;base64,{base64Image}")
            };

            var chatRequest = new ChatCompletionOptions();
            var chatClient = client.GetChatClient(deployment);

            var response = await chatClient.CompleteChatAsync(chatMessages, chatRequest); // Stuck here.

            var content = response.Value.Content;
        }
        catch(Exception ex) 
        {
            throw;
        }
    }

    private static string EncodeImage(string imagePath)
    {
        byte[] imageBytes = File.ReadAllBytes(imagePath);
        return Convert.ToBase64String(imageBytes);
    }
}

This code is getting stuck when CompleteChatAsync() is invoked. I waited for more than 5 minutes and there was no response.

When I tried the same code using python, it returned the response within 5-6 seconds.

from openai import AzureOpenAI  
import base64

endpoint = '/'
deployment = 'gpt-4o'
api_key = "LFK"
api_version = "2024-05-01-preview"

def encode_image(image_path):
    """Encodes an image to base64 format."""
    with open(image_path, "rb") as image_file:
        return base64.b64encode(image_file.read()).decode("utf-8")

def analyze_image(image_path):
    base64_image = encode_image(image_path)

    client = AzureOpenAI(
        azure_endpoint=endpoint,
        api_key=api_key,
        api_version=api_version
    )

    response = client.chatpletions.create(
        model="gpt-4o",
        messages=[
            {"role": "system", "content": "Analyze the uploaded image and return a single-word description of the main subject. The response should be only one word, representing the most general yet accurate category."},
            {
                "role": "user",
                "content": [
                   {"type": "text", "text": f"What ?"},  
                   {"type": "image_url", "image_url": {"url": f"data:image/png;base64,{base64_image}"}}
                ],
            }
        ]
    )

    return response.choices[0].message.content

image_path = "image3.png" 

result = analyze_image(image_path)
print("AI Response:", result)

Why is the .Net code not responding?

I have a program in .Net Core to describe an image using Azure OpenAPI GPT-4o model.

using Azure;
using Azure.AI.OpenAI;
using OpenAI;
using OpenAI.Chat;
using System;
using System.ClientModel;
using System.Collections.Generic;
using System.IO;
using System.Threading.Tasks;

// Azure OpenAI client library for .NET - version 2.1.0

public class AzureOpenAiService : IAzureOpenAiService
{
    private static string endpoint = "https://xyz.openai.azure.com/";
    private static string deployment = "gpt-4o";
    private static string apiKey = "LFK";

    public async Task FindPrimarySubjectAsync(string imagePath)
    {
        try
        {
            string base64Image = EncodeImage(imagePath);

            var credential = new AzureKeyCredential(apiKey);
            OpenAIClientOptions openAIClientOptions = new OpenAIClientOptions
            {
                Endpoint = new Uri(endpoint)
            };

            var client = new AzureOpenAIClient(new Uri(endpoint), new ApiKeyCredential(apiKey));

            var chatMessages = new List<ChatMessage>
            {
                new SystemChatMessage("Analyze the uploaded image and return a single-word description of the main subject. The response should be only one word, representing the most general yet accurate category."),
                new UserChatMessage($"What is in this image? Image: data:image/png;base64,{base64Image}")
            };

            var chatRequest = new ChatCompletionOptions();
            var chatClient = client.GetChatClient(deployment);

            var response = await chatClient.CompleteChatAsync(chatMessages, chatRequest); // Stuck here.

            var content = response.Value.Content;
        }
        catch(Exception ex) 
        {
            throw;
        }
    }

    private static string EncodeImage(string imagePath)
    {
        byte[] imageBytes = File.ReadAllBytes(imagePath);
        return Convert.ToBase64String(imageBytes);
    }
}

This code is getting stuck when CompleteChatAsync() is invoked. I waited for more than 5 minutes and there was no response.

When I tried the same code using python, it returned the response within 5-6 seconds.

from openai import AzureOpenAI  
import base64

endpoint = 'https://xyz.openai.azure.com/'
deployment = 'gpt-4o'
api_key = "LFK"
api_version = "2024-05-01-preview"

def encode_image(image_path):
    """Encodes an image to base64 format."""
    with open(image_path, "rb") as image_file:
        return base64.b64encode(image_file.read()).decode("utf-8")

def analyze_image(image_path):
    base64_image = encode_image(image_path)

    client = AzureOpenAI(
        azure_endpoint=endpoint,
        api_key=api_key,
        api_version=api_version
    )

    response = client.chat.completions.create(
        model="gpt-4o",
        messages=[
            {"role": "system", "content": "Analyze the uploaded image and return a single-word description of the main subject. The response should be only one word, representing the most general yet accurate category."},
            {
                "role": "user",
                "content": [
                   {"type": "text", "text": f"What ?"},  
                   {"type": "image_url", "image_url": {"url": f"data:image/png;base64,{base64_image}"}}
                ],
            }
        ]
    )

    return response.choices[0].message.content

image_path = "image3.png" 

result = analyze_image(image_path)
print("AI Response:", result)

Why is the .Net code not responding?

• python
• .net
• azure
• openai-api
• gpt-4o

1 Answer 1

I am able to resolve this issue.

The first thing to correct is that we have to use BinaryData instead of base64 encoded string.

byte[] imageBytes = File.ReadAllBytes(imagePath);
var binaryImage BinaryData.FromBytes(imageBytes);

The next thing is to correctly create UserChatMessage object to pass image in the chat.

 var chatImageContent = ChatMessageContentPart.CreateImagePart(binaryImage, "image/png");
 var chatTextContent = ChatMessageContentPart.CreateTextPart("What is in this image?");

 var userChatMessage = new UserChatMessage(chatTextContent, chatImageContent);

 var chatMessages = new List<ChatMessage>
 {
    new SystemChatMessage("Analyze the uploaded image."),
    new UserChatMessage(chatTextContent, chatImageContent)
 };

var chatClient = client.GetChatClient(deployment);

var response = await chatClient.CompleteChatAsync(chatMessages, chatRequest);

var content = response.Value.Content[0].Text;