Facial Attractiveness Prediction Using Vision Transformer

抄録

In recent years, many studies have been conducted on using deep learning to predict facial attractiveness. These studies are expected to have various applications such as face editing and beautification. Therefore, it is crucial to improve the prediction accuracy. In this study, I constructed a model for predicting facial attractiveness using the Vision Transformer, which has attracted much attention in the field of image recognition in recent years. The results show that the model improves the accuracy of facial attractiveness prediction compared to a simple convolutional neural network (CNN). By confirming the prediction factors in detail, this study is expected to contribute to our understanding of human perceptual characteristics and engineering applications.