Google Presents New Technology for Image Recognition

Google is currently researching a feature that would make identifying landmarks in your vacation photos a much simpler job. As part of Google's computer vision research program, the technology will automatically analyze the photos you submit and will match them with others from its growing database of pictures with geolocation information.

The new technology “that enables computers to quickly and efficiently identify images of more than 50,000 landmarks from all over the world with 80% accuracy” will be presented today at the Computer Vision and Pattern Recognition (CVPR) conference in Miami, Florida, by Google engineers. It is still in early research for now and a new product isn't planned yet, but, already, the results are promising.

“To be clear up front, this is a research paper, not a new Google product, but we still think it's cool. For our demonstration, we begin with an unnamed, untagged picture of a landmark, enter its web address into the recognition engine, and poof — the computer identifies and names it: ‘Recognized Landmark: Acropolis, Athens, Greece.’ Thanks computer,” the blog post announcing the technology reads.

The feature takes advantage of more than 40 million images with geolocation information that Google has access to in its Picasa and Panoramio services. Those pictures are used to create a list of well-known landmarks, which are then matched to known pictures of those landmarks. The software then uses an indexing system for image recognition to cluster all related pictures together.

When an image is submitted to be analyzed, the algorithm uses a pixel-matching technique to identify patterns in the picture and then compare them to those of known landmarks. The process works with a success rate of 80 percent and provides not only the name of the landmark, but its GPS data as well.