[Python] Any free non limited and fast local solution for extracting structured data from...

I want to build a program that extracts Data from invoices like TVA TTC TH the emitter the client etc... (mainly just these) also the invoices could be very different in structure to one another

I have tried doing OCR tesseract + LLM mistral but its just so slow on my PC (I want it to be relatively fast like 5 secs of execution even in normal PC specs) And tesseract sometimes struggles with scanned PDFs I tried to replace mistral by Gemini but its limited so I can't use that too. I also tried regex expressions after OCR but those are just not reliable.

Continue reading...