not that I've seen, though with some logical combination of object dimensions, env variables (depth of field, etc), and a decent image recognition/ML then you could probably get decent OCR. Note that this is for visual recognition, feel text is decently solved (exhibitA, Google Translate).