Geekspeak: April 30, 2001
California Institute of Technology researchers are teaching computers to notice and recognize human faces in scenes, not just in mug-shot poses but in arbitrary orientations.
This photo shows corner and circle feature identifications in a hallway display for the project trippingly titled "Viewpoint-Invariant Detection and Learning of Human Heads," fittingly found while I was taking a shortcut through CalTechs Moore Laboratory. The lab, a gift from Intel co-founder Gordon Moore, is generating plenty of work to occupy next-generation microprocessors.
For publications on this and other work by the CalTech Vision Group, see www.vision.caltech.edu.