I was trying to explain to someone new to search engines (yes these people do, apparently, still exist) how the various tools across Google work; predictive search, social, geo-targeted and so on.
This led into chats about Google Goggles at which point, his eyes glazed over and he seemed to stare through me. This reaction led me to believe it would be a great topic for my fellow geek readers.
Google Goggles is a downloadable image recognition application meaning through us, Google has eyes in every orifice and intimate corner we search for and from.
Snap happy researchers using the app send a collectively huge bank of images to a series of backend engines coordinated by Google’s Superroot Server which, identifies the source as either; text, geographic location, QR codes, corporate identities, products… you get the idea. All similar images are tagged and confidence scored* for future use.
As the database of images builds so will the use and accuracy of the app, making it easier to search for things that are difficult to summarise in text through traditional keyword search (wow, keyword search has been around so long now it feels normal to call it traditional- in web life anyway).
So the next time you’re out and about and you take a picture of a restaurant you want to dine in, you’ll know what happens behind the scenes to serve you the results that appear.
Simple really, thanks Google!
*Confidence scoring considers various weighted validation parameters such as quality and source, to determine the accuracy of the data held in the image.