Abstract
In recent years, image generation has made great strides in improving thequality of images, producing high-fidelity ones. Also, quite recently, thereare architecture designs, which enable GAN to unsupervisedly learn the semanticattributes represented in different layers. However, there is still a lack ofresearch on generating face images more consistent with human aesthetics. Basedon EigenGAN [He et al., ICCV 2021], we build the techniques of reinforcementlearning into the generator of EigenGAN. The agent tries to figure out how toalter the semantic attributes of the generated human faces towards morepreferable ones. To accomplish this, we trained an aesthetics scoring modelthat can conduct facial beauty prediction. We also can utilize this scoringmodel to analyze the correlation between face attributes and aesthetics scores.Empirically, using off-the-shelf techniques from reinforcement learning wouldnot work well. So instead, we present a new variant incorporating theingredients emerging in the reinforcement learning communities in recent years.Compared to the original generated images, the adjusted ones show cleardistinctions concerning various attributes. Experimental results using theMindSpore, show the effectiveness of the proposed method. Altered facial imagesare commonly more attractive, with significantly improved aesthetic levels.