We present a novel clustering method using HMM parameter space and eigenvector decomposition. Unlike the existing methods, our algorithm can cluster both constant and variable leng...
Automatic image orientation detection for natural images is a useful, yet challenging research area. Humans use scene context and semantic object recognition to identify the corre...
Abstract. This paper describes methods and sensor technology used to identify persons from their walking characteristics. We use an array of simple binary switch floor sensors to d...
—Facial expression is a natural and powerful means of human communication. Recognizing spontaneous facial actions, however, is very challenging due to subtle facial deformation, ...
Aggregating statistical representations of classes is an important task for current trends in scaling up learning and recognition, or for addressing them in distributed infrastruc...