Han et al [16] method is based on three stages: extract face
features, evaluate features and classification. They used 3
dimensional GavabDB erect of triangular network having no
makeup and jewelry etc. Geometric based method is used to
extract facial features. They place landmarks on face areas like
nose, mouth and eyes etc. They study the basic difference
between male and female and point out that shape of male
eyebrow is straight and also thick as compared to female while
female eyebrow is thinner. Female nose is smaller and have
small bridge as compared to male. Support vector machine
train those geometric features and efficiently classified gender.
They describe that error rate indication is 17.44% on average.
Their gender face database contained 427 images.