Interface 2003

A Boosted CCCD Classifier for Fast Face Detection
Diego Socolinsky, (Equinox Corporation),,
Joshua Neuheisel, (Equinox Corporation),,
Carey Priebe, (Johns Hopkins University),, and
Jason DeVinney, (Johns Hopkins University),


We introduce a fast object detection algorithm based on the Class-Cover Catch Digraph (CCCD) classifier. When applied to face detection, our algorithm exhibits excellent performance with speeds close to those of the fastest reported techniques. The main technical innovations in our method include a boosted tree-like CCCD classifier with a maximum rejection bias, and the use of a cross-correlation metric for fast similarity computation.

Take me back to the main conference page.