Like parent said, the patent doesn't actually seem have anything to do with facial recognition. I read it as talking about images that have already been identified, by some unspecified process. The process could be facial recognition, but it could also just be the user herself manually identifying each image.
I do agree that the patent is too unclear and broad, and that everything described must have been already known by 2008.
I'm not fluent in patent lawyer, so I must say you might be right. I read "subject identification" as an active process that was part of the claim, but it could refer to a pregenerated tag instead.
Which would make the patent even more laughable. "Wirelessly transfer all images I've tagged with 'Vacation 2017'" is patentable?
I do agree that the patent is too unclear and broad, and that everything described must have been already known by 2008.