Dual Encoder-Decoder Based Generative Adversarial Networks for Disentangled Facial Representation Learning

作     者：Hu, Cong Feng, Zhenhua Wu, Xiaojun Kittler, Josef 

作者机构：Jiangnan Univ Sch Artificial Intelligence & Comp Sci Wuxi 214122 Jiangsu Peoples R China Jiangnan Univ Jiangsu Prov Engn Lab Pattern Recognit & Computat Wuxi 214122 Jiangsu Peoples R China Minjiang Univ Fujian Prov Key Lab Informat Proc & Intelligent C Fuzhou 350121 Peoples R China Univ Surrey Dept Comp Sci Guildford GU2 7XH Surrey England Univ Surrey Ctr Vis Speech & Signal Proc Guildford GU2 7XH Surrey England 

出 版 物：《IEEE ACCESS》 (IEEE Access)

年 卷 期：2020年第8卷

页      面：130159-130171页

核心收录：

基　　金：Engineering and Physical Sciences Research Council (EPSRC) [EP/N007743/1] National Natural Science Foundation of China [U1836218, 61672265, 61876072, 61902153] 111 Project of Chinese Ministry of Education [B12018] Open Fund Project of Fujian Provincial Key Laboratory of Information Processing and Intelligent Control, Minjiang University [MJUKF-IPIC202002] EPSRC [EP/N007743/1, EP/R018456/1] Funding Source: UKRI 

主　　题：Face Gallium nitride Generative adversarial networks Training Generators Face recognition Task analysis Disentangled representation learning encoder-decoder generative adversarial networks face synthesis pose invariant face recognition 

摘      要：To learn disentangled representations of facial images, we present a Dual Encoder-Decoder based Generative Adversarial Network (DED-GAN). In the proposed method, both the generator and discriminator are designed with deep encoder-decoder architectures as their backbones. To be more specific, the encoder-decoder structured generator is used to learn a pose disentangled face representation, and the encoder-decoder structured discriminator is tasked to perform real/fake classification, face reconstruction, determining identity and estimating face pose. We further improve the proposed network architecture by minimizing the additional pixel-wise loss defined by the Wasserstein distance at the output of the discriminator so that the adversarial framework can be better trained. Additionally, we consider face pose variation to be continuous, rather than discrete in existing literature, to inject richer pose information into our model. The pose estimation task is formulated as a regression problem, which helps to disentangle identity information from pose variations. The proposed network is evaluated on the tasks of pose-invariant face recognition (PIFR) and face synthesis across poses. An extensive quantitative and qualitative evaluation carried out on several controlled and in-the-wild benchmarking datasets demonstrates the superiority of the proposed DED-GAN method over the state-of-the-art approaches.