Transferring Rich Deep Features for Facial Beauty Prediction

Feature extraction plays a significant part in computer vision tasks. In thispaper, we propose a method which transfers rich deep features from a pretrainedmodel on face verification task and feeds the features into Bayesian ridgeregression algorithm for facial beauty prediction. We leverage the deep neuralnetworks that extracts more abstract features from stacked layers. Throughsimple but effective feature fusion strategy, our method achieves improved orcomparable performance on SCUT-FBP dataset and ECCV HotOrNot dataset. Ourexperiments demonstrate the effectiveness of the proposed method and clarifythe inner interpretability of facial beauty perception.