HOG Based Facial Recognition Approach Using Viola Jones Algorithm and Extreme Learning Machine

Abstract

Extreme Learning Machine has attracted widespread attention for its exemplary performance in solving regression and classification problems. It is a type of single layer feed-forward neural machine which relies on randomly allocating the input weights and hidden layer biases. Through this, the ELM has been found to possess running time spans which are within millisecond regime. It does not require complex controlling parameters which makes its implementation elementary. This paper investigates the performance of employing Extreme Learning Machine as a classifier to be used for the face recognition problem. Viola Jones algorithm has been employed to detect and extract the faces from the dataset. Finally, Histogram of Oriented Gradients (HOG) features are extracted which form the basis of classification. The scheme so presented has been tested on standard face recognition datasets from AT&T and YALE. The resulting training/testing time spans of the whole scheme range from milliseconds to seconds, dictating the compatibility of ELM with real-time events.