Our research chiefly concerns the automated detection and description of humans in images, mainly by analysis of faces.
Members: John Robinson, Andy Pomfret, Justen Hyde, Matt Day, Jie Li, Ziran Wu
We have developed a new scheme for covariance matrix regularization that gives excellent results when applied to appearance-based face processing. Using conditional density estimation, we have shown how 3D face structure can be recovered from mugshots without explicit modeling of shape. We are investigating methods for speeding up the training phase of cascaded classifiers that use Haar basis functions for face detection. We have also done some work on the registration of 2D and 3D face information.