Google Wants to “See” the Web’s Trillion Images
July 2, 2008 by Jayvee Fernandez
Filed under Computers
Google currently indexes images primarily through text metadata. As anyone who’s used Google to search for porn pics knows, that technique doesn’t necessarily produce the best possible results. For the longest time, in fact, Google image search said I was a nightmare.
Google Image Labeler adds an extra social dimension to Google image search, but fun volunteer work cannot beat the ruthless efficiency of a bot for analyzing the Web’s one trillion images. That’s why Google is working on scene analysis and facial recognition — “visual search”, as Google’s RJ Pittman calls it in a recent interview.
A Googlebot with eyes — now that will bring us better porn.














