No question that there is still a lot of work to be done.
It's just a matter of time, not many years from now, until the machine will be able to classify objects in a image as best as we can; or better, considering that there are so many things we don't know about, thus we cannot recognize them (thinking of the actual animal, plant, rock formation, architecture type, location in the world, etc...)

What FaceBook's AI sees

What we see