Photo-realistic face age progression/regression using a single generative adversarial network.
Neurocomputing(2019)
摘要
•We formulate facial age synthesis as an unsupervised multi-domain image-to-image translation problem, and devise a novel generative framework using only a single generative adversarial network, dubbed FaceGAN which synthesizes photo-realistic face images with aging effects with unpaired samples and achieves face age progression and regression in a holistic framework.•In order to improve the capabilities of the generator and discriminator, we add an auxiliary multi-class classifier on top of the original discriminator and train the discriminator in a multi-task learning setting.•To further make aging effects of the synthesized face images more clear, a pre-trained deep face recognition model and a pre-trained deep age estimation model are introduced to preserve personal identity and age similarity respectively.•Experimental results show the superiority of our proposed method in terms of visual fidelity. We further empirically demonstrate the broad application capability of our approach on a facial attribute transfer and a facial expression synthesis tasks.
更多查看译文
关键词
Age progression/regression,Generative adversarial networks,Image-to-image translation
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络