[Demo paper] exploring attractive faces: General versus personal preferences

In this paper, we propose a novel Personality&Generality Support Vector Regression (PG-SVR) model to train the personality and generality regression attractiveness models from training facial images and their corresponding attractive scores simultaneously, which is completely different from existing method which returns only one general regression model. The trained PG-SVR serves for facial attractiveness enhancement, constructing low-dimensional reasonable solution space, which reflects the “Generality” and “Personality” attractiveness standard respectively. Experiments demonstrate that our PG-SVR enhanced face image space contains satisfactory results for different users and can be explored in real time.