R.J. Pittman, Google's Director of Product Management for Consumer Search Properties, was interviewed by Beet.tv. He suggests around 100 billion images are added online every year, and that Google expects over 1 trillion images online in the not-so-distant future. R.J. also talks about what Google calls "visual search" – actually "understanding" what's in the picture, not just analyzing the text in its vicinity. Like facial recognition, which could be used for identifiying people in an image, but also general object recognition...

http://blip.tv/scripts/flash/showplayer.swf?tabType3=none&tabUrl3=undefined&tabTitle3=undefined&tabType2=none&tabUrl2=undefined&tabTitle2=undefined&tabType1=none&tabUrl1=undefined&tabTitle1=undefined&enablejs=true&feedurl=http%3A%2F%2Fbeettv%2Eblip%2Etv%2Frss%2Fflash&file=http%3A%2F%2Fblip%2Etv%2Frss%2Fflash%2F1047864&thumb=http%3A%2F%2Fpanther2%2Evideo%2Eblip%2Etv%2FPlesstv%2DGoogleIsDevelopingImageProcessingToPowerImageSearchOfO719%2Epng&brandlink=http%3A%2F%2Fwww%2Ebeet%2Etv%2F&brandname=Beet%2ETV&showguidebutton=false&showplayerpath=http%3A%2F%2Fblip%2Etv%2Fscripts%2Fflash%2Fshowplayer%2Eswf

[Via http://searchengineland.com/080715-084529.php]

Check the slides from this page: http://google.client.shareholder.com/Visitors/event/build2/MediaPresentation.cfm?MediaID=31552&Player=1 (starting with slide #9).

Duh, guess it's old then. I still remember that those were the slides that didn't work during Google's press event.