Custom / Improve Line detection>

[lolsborn]

We often process legal documents with a fixed number of lines (usually 25) and would like to preserve the whole line. We could do this in a post-process, but I was wondering if there is a way to tweak line detection in doctr?

[felixdittrich92]

Hi @lolsborn 👋,

Excuse the late response.

You can pass paragraph_break=1000, to the ocr_predictor this high value raises the algorithm to keep everything as one line.

Hope this helps.

Best,
Felix

[lolsborn]

This worked wonders, thanks @felixdittrich92.

After: