earticle

영어

In modern life, we need better techniques based on biometric features recognition such as face recognition, fingerprint recognition and iris recognition. We present a method which can be used for face recognition or verification applications. The method can solve the problem that when the number of data categories is large and each number of the category used for training is small. As the conventional four stages, face detection, face alignment, face representation and face classification, we propose a Siamese architecture especially for the representation stage and use a one-against-one support vector machine for the classification stage. LFW dataset is used for training and testing which gets a considerable result. And we also test our system on other face dataset, which has a high accuracy on the recognition.