作者: Guodong Guo , Guowang Mu
DOI: 10.1109/CVPRW.2010.5543608
关键词: Gender variation 、 Scale (social sciences) 、 Statistics 、 Age and gender 、 Contextual image classification 、 Artificial intelligence 、 Ethnic group 、 Psychology 、 Estimation 、 Data mining 、 Facial recognition system
摘要: In this paper we study large-scale ethnicity estimation under variations of age and gender. The biologically-inspired features are applied to classification for the first time. Through a large number experiments on database with more than 21,000 face images, systematically effect gender estimation. Our finding is that can have high accuracy in most cases, but an interesting phenomenon observed ethnic accuracies could be reduced by 6∼8% average when female faces used training while males testing. results provide guide processing multi-ethnic database, e.g., image collection from Internet, may inspire further psychological studies grouping variations. We also apply methods whole MORPH-II 55,000 images five races. It time performed so database.