How to predict the global instantaneous feeling induced by a facial picture?

作者：

Highlights：

• A model of aesthetic quality assessment of frontal facial images based on low-level image statistics is proposed.

• It is shown that combining 4 learning algorithms (SVM, ANN, RF, and GBT) enhances the prediction performance and robustness.

• A model of likability evaluation based on high-level attributes is proposed.

• Likability and aesthetic quality estimations are combined to select automatically good quality images depicting likable faces.

摘要

Highlights•A model of aesthetic quality assessment of frontal facial images based on low-level image statistics is proposed.•It is shown that combining 4 learning algorithms (SVM, ANN, RF, and GBT) enhances the prediction performance and robustness.•A model of likability evaluation based on high-level attributes is proposed.•Likability and aesthetic quality estimations are combined to select automatically good quality images depicting likable faces.

论文关键词：Aesthetic quality,Likability,Automatic scoring,Portraits

论文评审过程：Available online 14 April 2015, Version of Record 3 December 2015.

论文官网地址：https://doi.org/10.1016/j.image.2015.04.002