Face detection based on hierarchical support vector machines
Yong Ma, Xiaoqing Ding · 2003
This paper presents a method of detecting faces based on hierarchical Support Vector Machines (SVM). The hierarchical SVM classifier is composed of a Combination of Linear SVM (CLSVM) and a nonlinear SVM. In training stage, the nonlinear SVM is trained under the constraint of the CLSVM to select more effective non-face samples. In detection stage, the CLSVM is used to fast exclude most non-faces in images and the nonlinear SVM is used to verify possible face candidates further. Experimental result on several databases demonstrates the feasibility of the method.