Good piece in yesterday's New York Times on this very topic, about a project at Princeton University:

http://www.nytimes.com/2012/11/20/science/for-web-images-creating-new-technology-to-seek-and-find.html

Of course, you have to be careful:

http://www.news.com.au/entertainment/television/petreaus-sex-scandal-tv-station-gets-caught-out-by-google-image-search 

Michele

-----Original Message-----
From: Code for Libraries [mailto:[log in to unmask]] On Behalf Of Joe Hourcle
Sent: Tuesday, November 20, 2012 7:10 PM
To: [log in to unmask]
Subject: Re: [CODE4LIB] Easiest way to tag thousands of images

On Nov 20, 2012, at 2:54 PM, Kyle Banerjee wrote:

> My real question is whether anyone has come up with a really good way 
> to assign metadata to thousands of photos, preferably in batch 
> fashion? Thanks,

There's a whole field of what they call 'computer vision', which is basically looking for things in images.  (face detection is just a small subset of it):

	http://en.wikipedia.org/wiki/Computer_vision

Most of the ones that I know of work with very constrained images (they're all pictures of the sun, at a known pixel scale, so they know what size/scale the items are that they're looking for)


The various 'find similar image' engines might be useful to do some sort of clustering of the images to make them easier to process.  (eg, extract landscapes vs. buildings vs. people)

Wikipedia has a list of various implementations:

	http://en.wikipedia.org/wiki/List_of_CBIR_engines

-Joe