Subjective experiments on gender and ethnicity recognition from different face representations

Abstract

The design of image-based soft-biometrics systems highly depends on the human factor analysis. How well can human do in gender/ethnicity recognition by looking at faces in different representations? How does human recognize gender/ethnicity? What factors affect the accuracy of gender/ethnicity recognition? The answers of these questions may inspire our design of computer-based automatic gender/ethnicity recognition algorithms. In this work, several subjective experiments are conducted to test the capability of human in gender/ethnicity recognition on different face representations, including 1D face silhouette, 2D face images and 3D face models. Our experimental results provide baselines and interesting inspirations for designing computer-based face gender/ethnicity recognition algorithms. © 2010 Springer-Verlag Berlin Heidelberg.