How to extract data from image using semantic kernal with AzureChatCompletion

[Aswin4070]

Hi everyone,

i am trying a project to extract data from image with the help of gpt4o.
I am able to do this with Azure Openai, but not able to do with semantic kernal.

Need help to develop this. When i tried to look at the documentation, i could see there is a library called ImageContent for dotnet:
https://github.com/microsoft/semantic-kernel/blob/main/dotnet/src/SemanticKernel.Abstractions/Contents/ImageContent.cs

but i cant see anything for python. So is there any other alternative available for python for data extraction from image?

[jamesmkfoo23]

I'm trying to do the same. How do i call SK plugins and pass the image (base64 or url) parameter?

[jamesmkfoo23]

https://youtu.be/ZaCjAKLMhYg?t=457
This seems to provide a way to add ImageContent to ChatMessageContentItemCollection in .NET. How about in Python?

[moonbox3]

Hi @Aswin4070, @jamesmkfoo23, we released the ability to do this in Python today (6/13/24) in v1.1.0. Our docs aren't updated to reflect this yet; however, you can find a concept sample showing how to work with this type of content.

[jamesmkfoo23]

Thanks @moonbox3 ! I got it working. Is there a away to specify the quality similar to how the REST API allows to?

[moonbox3]

Hi @jamesmkfoo23 apologies -- this property was missed and we will get it added early next week.

[jamesmkfoo23]

No apologies needed @moonbox3 ! Appreciate your speedy response and I'm truly impressed!
Thank you.