I am wondering if he runs any OCR or object recognition on collected screenshots.