I understand that. What I was asking is has anyone built custom nodes to do OCR recognition using available technologies