OCR basics: how text recognition works
Start here to separate character recognition, layout, confidence, text layers, and human review.
OCR guides for document accuracy, privacy and reviewable text
Use these guides to diagnose the earliest failing stage in an OCR workflow, choose an appropriate output, and verify the errors that would matter most.
Which OCR guide should you start with?
Begin with OCR basics when choosing an output or the OCR accuracy collection when diagnosing errors. Move to capture and preprocessing when pixels are weak, layout and structured data when position carries meaning, or languages and scripts for multilingual sources. The library also covers document workflows, privacy and safety, and archival quality control.
Choose a guide collection
Start with the problem you can see in the source, then follow the checks in order.
OCR accuracy guide to capture, diagnose and improve text
Use this lab to distinguish missing pixels, geometry, language, layout, and handwriting failures.
Document OCR workflows for PDFs, tables and forms
Plan searchable PDFs, tables, forms, receipts, and accessible documents around their distinct controls.
Online OCR privacy and safety guide for document uploads
Review authorization, minimization, retention, and safer offline routes for sensitive material.
Document capture and OCR preprocessing guide for better recognition
Use these capture tests when the source pixels, not the OCR engine, are the likely constraint.
OCR layout guide for tables, forms and structured data
Use this collection when correct characters still need reliable table, form, or page relationships.
Multilingual OCR guide for scripts and language models
Use these guides for multilingual, right-to-left, vertical, and diacritic-sensitive sources.
Archival OCR and quality control methods
Use this collection to design repeatable OCR quality controls for large or lasting collections.