A Neural Network Approach to Subjective Human Face Perception Classification based on Social Characteristics
Haojiang Ying, Yiyang Chen · 2021 IEEE 10th Data Driven Control and Learning Systems Conference (DDCLS) · 2021
Human can perceive rich information from a given face quickly and effortlessly. However, the detailed computational mechanism of face perception is still unclear. Psychologists have studied the face perception using psychophysics and neuroimaging methods for decades, but are still far from forming a converged view of classifying the perception of facial characteristics. One reason is that the facial characteristics are somewhat intertwined but the degree to which is not fully clarified. On the other hand, although having agreements on the association between facial traits and perceived social characteristics, people still perceive faces with a certain level subjectivity. In this study, the authors take the advantage of neural network to propose an algorithm for human face perception classification. The performance of this algorithm is evaluated based on the elements of a database consisting of human face social characteristics, and the experimental results confirm its classification accuracy.