from :http://make3d.stanford.edu/index.html
Make3D converts your single picture into a 3-D model, completely automatically.
It takes a two-dimensional image and creates a three-dimensional "fly around" model, giving the viewers access to the scene's depth and a range of points of view. After uploading your image, you can "fly" in the 3-D scene (requires VRML viewer or Adobe Shockwave), or watch a rendered 3-d movie (flash required). Visit 3-D Gallery or Hall of fame.
How does it work?
It uses powerful machine learning techniques to learn the 3-d structure of a scene as a function of the (single) image features. The team led by Ashutosh Saxena and Andrew Y. Ng has made the code available online; see more details here.