Facial landmark disentangled network with variational autoencoder

作     者：LIANG Sen ZHOU Zhi-ze GUO Yu-dong GAO Xuan ZHANG Ju-yong BAO Hu-jun LIANG Sen;ZHOU Zhi-ze;GUO Yu-dong;GAO Xuan;ZHANG Ju-yong;BAO Hu-jun

作者机构：CAD&CG State LabsZhejiang UniversityHangzhou 310027China School of Mathematical SciencesUniversity of Science and Technology of ChinaHefei 230026China. 

出 版 物：《Applied Mathematics(A Journal of Chinese Universities)》 (高校应用数学学报（英文版）（B辑）)

年 卷 期：2022年第37卷第2期

页      面：290-305页

核心收录：

学科分类：12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 081104[工学-模式识别与智能系统] 08[工学] 0835[工学-软件工程] 0701[理学-数学] 0811[工学-控制科学与工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：Supported by the National Natural Science Foundation of China(61210007) 

主　　题：disentanglement representation deep learning facial landmarks variational autoencoder 

摘      要：Learning disentangled representation of data is a key problem in deep ***,disentangling 2D facial landmarks into different factors(e.g.,identity and expression)is widely used in the applications of face reconstruction,face reenactment and talking head et al..However,due to the sparsity of landmarks and the lack of accurate labels for the factors,it is hard to learn the disentangled representation of *** address these problem,we propose a simple and effective model named FLD-VAE to disentangle arbitrary facial landmarks into identity and expression latent representations,which is based on a Variational Autoencoder ***,we propose three invariant loss functions in both latent and data levels to constrain the invariance of representations during training ***,we implement an identity preservation loss to further enhance the representation ability of identity *** the best of our knowledge,this is the first work to end-to-end disentangle identity and expression factors simultaneously from one single facial landmark.