This paper presents a new approach for the binarization of seriously degraded manuscript. We introduce a new technique based on a Markov Random Field (MRF) model of the document. ...
One commonly used approach for language recognition is to convert the input speech into a sequence of tokens such as words or phones and then to use these token sequences to deter...
Knowledge transfer is computationally challenging, due in part to the curse of dimensionality, compounded by source and target domains expressed using different features (e.g., do...
Boosting has become very popular in computer vision, showing impressive performance in detection and recognition tasks. Mainly off-line training methods have been used, which impl...
Several sound classifiers have been developed throughout the years. The accuracy provided by these classifiers is influenced by the features they use and the classification method ...