Nice. Are you using the Tessaract OCR lib at the core of the extraction?