A novel method for evaluating facial attractiveness

Beauty is an abstract concept that is inherently difficult to quantify and evaluate. The analysis of facial attractiveness has received much research attention in the past. Recent work has shown that facial attractiveness can be learned by machine, using supervised learning techniques. This paper proposes a computational method for estimating facial attractiveness based on Gabor features and support vector machine (SVM). We conducted several experiments using different feature types including Gabor features, geometric features, and eigenfaces. We found that the Gabor feature-based method produced the best result. To further improve the performance of this predictor, we combined Gabor features with geometric facial features, and a high correlation of 0.93 with average human ratings was achieved. This result indicates that our new approach performs well in the evaluation of facial attractiveness.