Simulation of Facial Palsy using Conditional Generative Adversarial Networks and Face Shape Normalization
Takato Sakai, Masataka Seo, Naoki Matsushiro, Yen‐Wei Chen · 2021 IEEE 10th Global Conference on Consumer Electronics (GCCE) · 2021
The Yanagihara method for evaluating facial nerve palsy is based on physicians’ visual examinations, and the problem with this approach is that it is difficult to use patients’ images for physicians’ training and educations because of patients’ privacy protection. We aimed to solve this problem by creating a pseudo-facial image of a patient with facial paralysis that can be shared among doctors. To reproduce the patient’s facial expression in a public face image, we proposed a method to generate a target face image using a generative adversarial model based on deep learning. In addition, we conducted an experiment to reproduce the face image while considering the difference in human face shapes. Based on experimental results, we successfully generated the target image. Furthermore, the data preprocessing method needed to be improved to reproduce the fine details of the face.