This is called Image Mining, and was heavily pioneered by Quaero.
Re: Why doesn’t Google release this?
Performance, performance, performance. Comparing images is several orders more difficult than processing text, as the process requires some fuzyy logic. E.g. it should find images which are essentially the same, but of different size or with different JPEG compression settings.
Marcin – Oh yeah, 'cause Google always shies away when the going gets tough! :-D
Google has no reason not to have something like this after their acquisition of neven vision.
Google has already said it will add this feature (and many other enhancements) in Image Search.
well then, there is no reason it should take them this long after their acquisition! but i guess there has to be *some* reason since it has taken them this long. i'm just impatient
I'm not saying it's impossible, but that it requires massive amounts of computing == lots of electricity == lots of $$$. Of course Google does have the money but finding locations for large datacenters with access to good connectivity, power and cooling is not so trivial. Clearly, they still have other projects with higher priority, than throwing tens of thousands servers just to launch the image search with pattern recognition.