We present a holistic statistical model for the automatic analysis of complex scenes. Here, holistic refers to an integrated approach that does not take local decisions about segme...
Daniel Keysers, Michael Motter, Thomas Deselaers, ...
Deformations in handwritten characters have classdependent tendencies. For example, characters of class “A” are often deformed by global slant transformation and never deforme...
Seiichi Uchida, Mohammad Asad Ronee, Hiroaki Sakoe
This paper presents a system which learns from examples to automatically recognize people and estimate their poses in image sequences with the potential application to daily surve...
We present a maximally streamlined approach to learning HMM-based acoustic models for automatic speech recognition. In our approach, an initial monophone HMM is iteratively refin...
Speech enhancement is a common approach to address the effects of degradation due to noise and channel contamination. This approach is intended to suppress unwanted signal and rec...