Google digs deeper into Image Recognition

23 Jun
2009

google_logo1Google scientists announced a new technology today that might bring big improvements to how Google processes and understands images. For now, it’s limited to images of landmarks like the Golden Gate Bridge and the Eiffel Tower, but conceptually, at least, its promise is much broader.

Jay Yagnik, Head of Computer Vision Research said:

To be clear up front, this is a research paper, not a new Google product, but we still think it’s cool,

This is no easy task when the engine has to rely on images of the landmarks, which are incredibly varied by angle, lighting, photo quality, etc.

According to Google it has managed to achieve 80% of the total accuracy on more than 50,000 landmarks. Google demonstrated that how it did this with the Acropolis. Google began with an unnamed, untagged picture of it, entered the web address into the recognition engine, and the computer identified it as “Recognized Landmark: Acropolis, Athens, Greece.”

Acropolis Cluster

Google generated a list of landmarks based on GPS-tagged photos from Picasa and Panoramio, and online tour guide webpages. Google then found “candidate images” for each landmark using those resources as well as Google Image Search.

Images were “pruned using efficient image matching and unsupervised clustering techniques, Yagnik explained.

Google said in that research paper:

Here, we build a world-scale landmark recognition engine, which organizes, models and recognizes the landmarks on the scale of the entire planet Earth, constructing such an engine is, in essence, a multi-source and multi-modal data mining task.


Comment Form