Using docling for images

[Mar-Lourenco]

Hello! I have a pdf file that has images and I would like to extract the content of that image. Do you know how I can do that with docling? Thanks in advance.

[ramiroortiz]

Please review the following link https://ds4sd.github.io/docling/examples/export_figures/ that shows how to export the images in a markdown file. If you want to process the images you can use a visual llm like llama 3.2 vision to interpret the content of the images.

[Mar-Lourenco]

Thank you. With the DocumentConverter() and export_to_dict() I am able to remove the picture coordinates. My idea is to use them to be able to extract the text inside the picture. Do you know how I can do that with docling or another tool?

[kiril-buga]

You can extract text from images by setting pipeline_options.do_ocr = True