Google Blogoscoped

Saturday, June 17, 2006

Turning 2D Into 3D

Carnegie Mellon University researchers have found ways to construct a 3D scene out of a flat 2D image. This could be useful for robots moving around, security cams or archive photos.

The 300 images to train the program came from a Google search, the article says; based on those “examples of vertical and horizontal surfaces” a machine learning program was able to “develop statistical associations between certain shapes, shadings and other characteristics typical of each orientation.”

For example, when you can determine a shape is a car or a chair you know it must be positioned on a flat surface. Of course, determining that a bunch of pixels make up a car or chair is an immense task for traditional algorithms (it’s one of those things CHI might be good for).

[Via Digg. Image from the Carnegie Mellon Press Release, June 13 2006.]


Blog  |  Forum     more >> Archive | Feed | Google's blogs | About


This site unofficially covers Google™ and more with some rights reserved. Join our forum!