in the interest of understanding, is there any code or similar for the approach? does that OCR run anywhere today?