Efficient Facial Attractiveness Prediction Using Dual Label Distribution and Lightweight Design

A novel end-to-end facial attractiveness prediction approach that integrates dual label distribution and lightweight design to achieve promising results with greatly decreased parameters and computations.