[Python] How to take a png of a page and split it into pngs of paragraphs in python

I have this dictionary I am trying to digitize, I have now split the pdf into individual pages but what I need to do now is to take each page's PNG and split it into many pictures of each individual paragraph/word.

I had a code that worked relatively well that used contours but I am stupid and forgot to commit the changes and I started trying other stuff and lost it. I have tried prompting AI with various stuff like using the spacing between lines and paragraphs or contours and none of it have worked. I saw some other examples online like this other question, the structure they used was a bit different so I tried to change it but didn't work either as it just gives me the entire page. The best I have gotten is it gets some words but they are completely random and not individual paragraphs.

Continue reading...