Smartphone users no longer need to think up an appropriate phrase when searching the web on the go 鈥� at least, not if they are using Google鈥檚 Android operating system.
鈥淵ou take a picture of an item and use that picture as the query,鈥� explains , one of Google鈥檚 vice presidents for engineering.
The service, called , attempts to identify the object 鈥� for example, it might recognise a landmark like the Golden Gate bridge in San Francisco 鈥� and returns a page of conventional search results, as if a user had typed the name of the landmark. Goggles can also identify artworks and extract the text from a photo, for example of a business card. It might find places to buy a product you鈥檝e seen or identify local attractions.
Advertisement
When a user takes a photograph, Goggles sends the image to a Google server where it is analysed by algorithms looking for 鈥渟ignatures鈥� of objects within the image, says Gundotra. The signature is then compared against a database of more than a billion images.
鈥淭he best matches are ranked and sent back down to your device, all in a fraction of a second,鈥� says Gundotra.
Point, shoot, buy
Goggles is not the first visual search engine, nor is it the first available as a mobile app. Since 2006 , the largest cellphone applications provider in Japan, has enabled users to buy products such as CDs by taking photos of them. The system uses software developed by in Pasadena, California.
The usefulness of searching visually was always going to come to Google鈥檚 attention, says Paolo Pirjanian, Evolution Robotics鈥� president and CEO.
鈥淚t makes it possible to do a lot more than you can describing something by words alone,鈥� he says. 鈥淚t is an obvious next step for search.鈥�
A mouse pointer for the world
Google鈥檚 product is tailored to work best on certain types of object, such as landmark buildings, says Gundotra, but the goal is to identify any object in any image. 鈥淚t will be as simple and easy as pointing at an object. You will be able to treat it like a mouse pointer for the real world.鈥�
However, Google has set itself some limits, to fend off privacy fears.
鈥淲hile we have the underlying computer science technology to do facial recognition, we have decided to delay that,鈥� Gundotra says, citing privacy concerns. Google acquired face-recognition technology developed by Neven Vision in 2006.
