document intelligence figure detection

[chwunderpers]

Is your feature request related to a problem? Please describe.
You got the analyzed content from Document Intelligence of your document in markdown. You also got a list of detected figures from Document Intelligence.

The number of detected figures in the content

... don´t match the number of figures in the list of detected figures from Document Intelligence!

The problem is, how to "link"

to the appropriate figure from the figures list?

In my case, I would like to replace the images in the text with an LLM generated text for those images. This would need to be placed at the same location than the original image.

How do I do this, if number of

found in content is 12 but the number of figures from DI is 13?

Describe the solution you'd like
instead of just writing

in the markdown analyzed content, you could add the figure id in this. like <figure 15.1> or simething like this, so you can easily identify which detected images related to the .

[xiangyan99]

Thanks for the feedback, we’ll investigate asap.

[github-actions]

Thanks for the feedback! We are routing this to the appropriate team for follow-up. cc @bojunehsu @vkurpad.

[rehanga937]

I'm glad I found this thread; I have a very similar use case as well. A total of 80 images are detected in my document, however the markdown only mentions 60.

[bojunehsu]

@chwunderpers, @rehanga937 Can you please share the original document and result JSON with us via formrecog_contact@microsoft.com so we can investigate? Thanks.