Joan Capell Gracia

Hello, I'm not sure if I can help you with the implementation but in my option the easier and best solution is to crop the box that has the higher...

It won't correct the boxes ("Hellow", "oworld") case but it will at least solve correctly the rest

Yes nb_chars = the predicted number of characters Lets say that we have the words "hello world" but the ocr recognize "hello oworld". You have two bounding boxes with same...

Yeah you're right that my proposition will be random in too many cases, using an histogram will probably be better.

It's the same for me

If it is a large pdf file it also needs to be chunked in and then perform a query to only input the relevant chunks into the LLM as context.

Could you give me some hints as to how could I create a plugin with such functionality?

I have the same problem

I used paperspace gradient with a P500