Ask for function ?

[chenboju]

Hello, is this function available?
When converting pdf to markdown, we hope to extract both images and tables. For image extraction, we hope to be able to use LLM to view the image, give a description of the image through the context, and have it appear in the markdown content.

[github-actions]

👋 Welcome! Thanks for opening your first issue. If you'd like to take a crack at fixing it, feel free to open a pull request — otherwise, we'll take a look as soon as we can!

[mandalsouvik3333]

No, it's not there. But it should be a simple task to add. Let me know if you are interested in adding this, I can help you

[chenboju]

Thank you very much. I have tried many OCR methods, but they all refer to extracting images without explaining the content of the images.
Currently, only marker is known to work, while mineru needs to be processed twice.

[souvik3333]

@chenboju This is added. We have also open sourced a model for this task: https://huggingface.co/nanonets/Nanonets-OCR-s

[chenboju]

Thanks

[mandalsouvik3333]

closing this