Make a Face: Towards Arbitrary High Fidelity Face Manipulation

Qian, Shengju; Lin, Kwan-Yee; Wu, Wayne; Liu, Yangxiaokang; Wang, Quan; Shen, Fumin; Qian, Chen; He, Ran

doi:10.1109/iccv.2019.01013

Cited by 81 publications

(56 citation statements)

References 46 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Cole et al [26] decomposed face into a set of sparse landmark points and aligned texture maps, which are combined to generate front faces by image warping. Qian et al proposed Additive Focal Variational Auto-encoder (AF-VAE) [27] that combined VAE with GAN, introducing a new random gaussian mixture hypothesis to improve the de-entanglement effect of face content and obtain realistic frontal images. Kan et al [28] proposed Stacked Progressive Auto-encoder (SPAE), which iteratively transforms large poses to virtual smaller pose until target pose.…”

Section: Related Workmentioning

confidence: 99%

Detailed Feature Guided Generative Adversarial Pose Reconstruction Network

Hao

Chen

2021

IEEE Access

View full text Add to dashboard Cite

Face frontalization is a critical and difficult task to face pose reconstruction. Previous researches use simple posture information as guidance, such as pose coding and facial landmarks. To explore the guidance effect of profile faces, we propose detailed features that provide much detailed information. In this paper, we devise Detailed Feature Guided Generative Adversarial Pose Reconstruction Network (DGPR). Firstly, frontal pose coding and profile detailed features are fed into DGPR to generate detailed features of front face. Then, the second generator combines frontal detailed features and profile face to obtain front face. Besides, we propose a conditional enhancement loss to strengthen the guiding role of detailed features, and a smoothing loss to reduce edge sharpness in generated faces. Experimental results show that our method generates photorealistic front faces and outperforms state-of-the-art methods on M 2 FPA and CAS-PEAL. Specifically, DGPR improves the face recognition accuracy under ±60 • , ±75 • , ±90 • by 2%, 1%, and 6% respectively, compared with state-of-the-art methods on M 2 FPA. And the average rank-1 recognition rate achieved 99.95% on CAS-PEAL, which is improved by 0.05%. The results demonstrate the effect of detailed features and corresponding modules.

show abstract

Section: Related Workmentioning

confidence: 99%

Detailed Feature Guided Generative Adversarial Pose Reconstruction Network

Hao

Chen

2021

IEEE Access

View full text Add to dashboard Cite

show abstract

“…ExprGAN [15] introduces a model which can control the intensity of the generated expressions conditioned on the discrete expression labels. [16] proposes Additive Focal Variational Autoencoder to manipulate expression where landmarks boundary maps are needed. GANimation [4] adopts AUs as expression labels and can generate expressions continuously.…”

Section: Related Workmentioning

confidence: 99%

Attention Based Facial Expression Manipulation

Wang

Xiang

Liu

et al. 2021

2021 IEEE International Conference on Multimedia &Amp; Expo Workshops (ICMEW)

View full text Add to dashboard Cite

Facial expression manipulation has two objectives: 1) generating an image with target expression; 2) preserving the identity information of the original image as much as possible. Recently, Generative Adversarial Networks (GANs) have shown the abilities for fine-grained facial expression manipulation. However, current methods are still prone to generate images with poor quality. In this work, we propose a U-Net based generator with multi-attention gate for facial expression manipulation. The multi-level attention mechanism is helpful to manipulate relevant regions and preserve identity features, thus improving the editing ability. Furthermore, we adopt self-attention block to replace direct skip-connection to get long-range dependency in images. To suppress artifacts in generated images, we add a discriminator based loss function in the training process. Extensive experiments on both quantitative and qualitative evaluation show that our proposed method achieves better performance for facial expression manipulation.

show abstract

“…针对高分辨率人脸图像的属性编辑问题, 上述方法由于缺乏充足的训练数据, 往往难以取得令人满意的效果. 为此, 商汤科技研究团队 [25] 提出一种可加性焦距变分自编码器, 通过对图像重建和相对熵损失进行弱监督训练, 实现高分辨率人脸图像的属性修改. 商汤科技与香港中文大学 [26] [27] 联合提出一种基于示例图像指导的图像属性编辑方法 PuppleGAN.…”

Section: 人脸编辑unclassified

Deep visual identity forgery and detection

彭春蕾¹,

高新波²,

王楠楠³

et al. 2021

Sci. Sin.-Inf.

View full text Add to dashboard Cite

With the wide application of deep learning techniques in video and image generation, the quality of visual identity forgery, especially face forgery, is becoming increasingly high. The detection of visual identity forgery has been a hot issue because of its important influence on both national security and social stability. In this paper, we introduce recent researches on deep visual identity forgery from target-specific face forgery and target-generic face forgery. We further summarize the techniques of detecting visual identity forgery from multiple categories, including the spatial clue based methods, the temporal clue based method, the techniques for generalizable forgery detection and spoofing forgery detection models. We later present the public datasets and performance of representative approaches on these datasets. Finally, the issues and challenges in existing research are discussed.

show abstract

Make a Face: Towards Arbitrary High Fidelity Face Manipulation

Cited by 81 publications

References 46 publications

Detailed Feature Guided Generative Adversarial Pose Reconstruction Network

Detailed Feature Guided Generative Adversarial Pose Reconstruction Network

Attention Based Facial Expression Manipulation

Deep visual identity forgery and detection

Contact Info

Product

Resources

About