Goggles, which is only available to Android handsets right now, allows users to take a picture of something and conduct a search based on the image. The image is sent to Google's servers where it is compared against other images with similar appearance. Once Google makes a match, it sends the results back to the phone. The service was first introduced in late 2009.
Goggles is good at some types of search, but not others. For example, taking a picture of a Porsche and sending it in for analysis isn't necessarily going to return information about the 911 Turbo S you've been eyeing. Instead, it is better with much more specific images. Petrou demonstrated that after taking a picture of a beer can, Goggles was able to correctly determine that the brand of the beer was Boddingtons (a fine choice, I must say).
While Google continues to refine and improve the software, it is also making sure to bring the product to Android's chief competitor, the iPhone.
According to Petrou, Google is weighing whether or not to open up to third-party developers. The problem is that Goggles requires control over the camera functions, including aspects such as white balance, exposure and more. This means Goggles needs to work independently as its own application, and not within the confines of another app. Still, Google is exploring new territory, and even suggested that it might marry the camera of smartphones with Goggles, which could lead to augmented reality visual search.
I can't say that I use Goggles too much, but it certainly has its appeal. I still think the best use for it is as a tourist who's exploring a new city for the first time, sees something unique, and wants to know what it is.