Anyone else surprised that this is a recent advancement, or is it just me? Perhaps I'm underestimating the difficulty of the implementation? But it seems to me

Anyone else surprised that this is a recent advancement, or is it just me? Perhaps I'm underestimating the difficulty of the implementation? But it seems to me anyone with any ability for introspection should have known that: When the human eye searches for an object it looks globally for the rough location, size and orientation of the object. Then it zeros in on the details