After waking up too early one morning (kids !) I had this idea. I nearly got up at 04:30 to start writing this. But I wasn't that daft.
Imagine a new functionality where you could search images for content. Starting with face recognition.
Upload a photo of the person you're looking for and Google could scan it for biometrics and look for that face in the image database (starting with Picasaweb ?). Optionally, add tags to refine the search.
Using Google's computing power they could scan existing images for faces and calculate all the relevant biometrics.
Using Google's search technology they could gather the feedback from people searching to refine the search results.
Once enough people start using it, the combination of information technology and human feedback would soon start to generate real context to images.
But damn ! I'm blogging on something (similar) which already exists; check out www.myheritage.com/celebrity-face-recognition. (Or specifically, why I look like Robert de Niro).
Well, I still expect it would be worth Google's while to expand their search market.
And heck, why just stop with faces ? How about buildings, flowers, trees, animals ...
All that context to be added ...
