What libraries do you see as being SOTA? Fitz? Tika?
My hope is that computer vision + OCR will solve this once and for all in near future.
My hope is that computer vision + OCR will solve this once and for all in near future.
Your second comment rings true, and in my opinion, we are there. Highly recommend throwing some PDFs at AWS Textract and checking out the quality, it wasn't there a few years ago, can safely state it's there now though. I threw stuff at it that previously would just spit out trash, and it handled it fairly well, specifically for table data extraction (I was looking at public stock market quarterly reports).
Cost is the kicker for me, 1000 pages for $15, adds up fairly quickly at any sort of scale!
I really want to see OCR become easier to use, but I don't know why it's such a hard problem in the first place.
I believe it uses tesseract, ghostscript and some other libraries.
Speaking of ghostscript, one way to deal with problematic PDFs is to print them to file and deal with the result instead.
I'd love to just be able to search a PDF document for a string and get a list of results.