Face Recognition using Gabor features and Adaptive resonance theory (ART)

Abstract

In this proposed contribution, we derive Gabor features of the face images both while training and testing experiments. Later we use a ARTMAP for classification that merges two somewhat customized ART-1 or ART-2 units into a supervised learning structure wherein the 1st unit accepts the input data (training set of images) and the 2nd unit acquires the accurate output data (image labels), after that are used to create the lowest possible adjustment of the vigilance parameter in the 1st unit so as to make the accurate classification. Our results of experiments show that performance comparable with the computationally intensive evolutionary methods could be achieved in much less time.