TY - JOUR
T1 - Deep generative image priors for semantic face manipulation
AU - Hou, Xianxu
AU - Shen, Linlin
AU - Ming, Zhong
AU - Qiu, Guoping
N1 - Publisher Copyright:
© 2023 Elsevier Ltd
PY - 2023/7
Y1 - 2023/7
N2 - Previous works on generative adversarial networks (GANs) mainly focus on how to synthesize high-fidelity images. In this paper, we present a framework to leverage the knowledge learned by GANs for semantic face manipulation. In particular, we propose to control the semantics of synthesized faces by adapting the latent codes with an attribute prediction model. Moreover, in order to achieve a more accurate estimation of different facial attributes, we propose to pretrain the attribute prediction model by inverting the synthesized face images back to the GAN latent space. As a result, our method explicitly considers the semantics encoded in the latent space of a pretrained GAN and is able to faithfully edit various attributes like eyeglasses, smiling, bald, age, mustache and gender for high-resolution face images. Extensive experiments show that our method has superior performance compared to state of the art for both face attribute prediction and semantic face manipulation.
AB - Previous works on generative adversarial networks (GANs) mainly focus on how to synthesize high-fidelity images. In this paper, we present a framework to leverage the knowledge learned by GANs for semantic face manipulation. In particular, we propose to control the semantics of synthesized faces by adapting the latent codes with an attribute prediction model. Moreover, in order to achieve a more accurate estimation of different facial attributes, we propose to pretrain the attribute prediction model by inverting the synthesized face images back to the GAN latent space. As a result, our method explicitly considers the semantics encoded in the latent space of a pretrained GAN and is able to faithfully edit various attributes like eyeglasses, smiling, bald, age, mustache and gender for high-resolution face images. Extensive experiments show that our method has superior performance compared to state of the art for both face attribute prediction and semantic face manipulation.
KW - Face attribute prediction
KW - GANs
KW - Semantic face manipulation
UR - http://www.scopus.com/inward/record.url?scp=85149616506&partnerID=8YFLogxK
U2 - 10.1016/j.patcog.2023.109477
DO - 10.1016/j.patcog.2023.109477
M3 - Article
AN - SCOPUS:85149616506
SN - 0031-3203
VL - 139
JO - Pattern Recognition
JF - Pattern Recognition
M1 - 109477
ER -