Comment by yorwba

What would you say is currently the most accurate OCR solution if you're not concerned about speed and memory usage?