Best way to handle OCR for scanned PDFs in a web app (cost vs accuracy)?

via

Python

Hey, I’m building a project where users upload PDFs and I need to extract text from them. For normal text PDFs, extraction works fine. But for scanned/image-based PDFs, I’m using Tesseract + some preprocessing. The problem is: Accuracy is inconsistent (especially on low-quality scans) Output needs cleanup Doesn’t handle structure well (tables, formatting, etc.) I’ve also looked into Google Vision OCR, but: It asks for card details (which is fine, but I’m cautious) Free tier is limited Not sure i