Automatical gender detection for unconstrained video sequences based on collaborative representation
Lijia Lu, Weiyang Liu, Yandong Wen, Yuexian Zou · 2014
Many intelligent systems are required to deal with the situation of human-computer interaction. As one of the most important front ends, gender classification plays an irreplaceable role. For practical use, a real-time robust gender classification system is presented in this paper. The system consists of three principal modules: image preprocessing, face detector and gender classifier. To enhance the classification accuracy with affordable complexity, Haar-like features and Ada-Boost-trained classifier are applied to the face detector while Eigenface features and collaborative representation classifier are embedded to the gender classifier. Experimental results verify the real-time ability and gender classification accuracy of the proposed system. It is worth mentioning that the system performs well when handling faces with occlusion and complex background.