A new approach to the recognition of temporal behaviors and activities is presented. The fundamental idea, inspired by work in speech recognition, is to divide the inference probl...
This paper proposes a method of automated individualization of eye region model. The eye region model has been proposed in past research that parameterizes both the structure and ...
This paper introduces a new concept of surveillance, namely, audio-visual data integration for background modelling. Actually, visual data acquired by a fixed camera can be easily ...
We present a novel method for inferring three-dimensional shape from a collection of defocused images. It is based on the observation that defocused images are the null-space of ce...
Abstract. The lack of eye contact in desktop video teleconferencing substantially reduces the effectiveness of video contents. While expensive and bulky hardware is available on th...