The Active Perception Laboratory (APL) is an interdisciplinary research center in the School of Computer Science at Carnegie Mellon and the Center for the Neural Basis of Cognition -- a collaboration between Carnegie Mellon and the University of Pittsburgh. The primary mission of the APL is to discover and elucidate the organizing computational principles and the neural basis of vision by integrating concepts and techniques from physics, mathematics, computer science and electrophysiology.

Director: Dr. Tai Sing Lee,
Associate Professor of Computer Science and Neural Basis of Cognition.


Mellon Institute 115
Carnegie Mellon University
4400 Fifth Avenue
Pittsburgh, PA 15213

Lab News:

Using our database of natural range (depth) and intensity images, we have developed an algorithm for enhancing low-resolution 3D shape estimates by using a full-resolution intensity image (see NIPS 2006). This is useful for cleaning up stereo output or inexpensive range scans. Our algorithm was twice as effective at reducing sum-squared error than Shape Recipes, a previous leading approach.

In this paper, we also show that in natural scenes, even in sunny outdoor conditions, shadow cues may be stronger for inferring depth than classical shading cues. This finding is related to our previous discovery that darker pixels in an image tend to be farther away (see JOSA 2003), which is related to the psychophysical depth cue of relative brightness:

Among bodies equal in size and distance, that which shines the more brightly seems to the eye nearer.
 - Leonardo da Vinci




Computational Neuroscience research at Carnegie Mellon.