There's no world in which those are close, because the former requires mapping pixels to numbers and the latter inferring meaning from prose.