Quantitative Evaluation of the Whitening Effects of Facial Masks via EfficientNet and the Wavelet Transform

As important identity information, faces have been widely considered. With the rapid development of the social economy and beauty industry, facial skin conditions have attracted increasing attention. As an important feature of faces, skin color and its categories and changes have also attracted widespread attention. In this work, Asians are divided into three categories according to their skin color characteristics: yellow, dark yellow and light yellow. The network model combining EfficientNet-B0 and the two-dimensional discrete wavelet transform is used as the classifier, which successfully classifies the skin color and improves the accuracy of the training results. On the basis of these results, the skin color is quantified, and an evaluation model of the facial mask whitening effect is designed. Given that the facial mask does not change obviously in a short time and that the evaluation period is too long, an evaluation model that can reflect the whitening and brightening effects of the facial mask in real time is designed. The brightening effect of the facial mask can be evaluated instantly and intuitively through changes in value, which provides effective suggestions for the choice of facial mask with different skin colors and has a wide range of practical application value and significance.