AttGAN: Facial Attribute Editing by Only Changing What You Want

By Zhenliang He, Wangmeng Zuo, Meina Kan, Shiguang Shan, Xilin Chen. 2017

828 citations recorded by the source corpus.

Facial attribute editing aims to manipulate single or multiple attributes on a given face image, i.e., to generate a new face image with desired attributes while preserving other details. Recently, the generative adversarial net (GAN) and encoder–decoder architecture are usually incorporated to handle this task with promising results. Based on the encoder–decoder architecture, facial attribute editing is achieved by decoding the latent representation of a given face conditioned on the desired attributes. Some existing methods attempt to establish an attribute-independent latent representation for further attribute editing. However, such attribute-independent constraint on the latent representation is excessive because it restricts the capacity of the latent representation and may result in information loss, leading to over-smooth or distorted generation. Instead of imposing constraints on the latent representation, in this work, we propose to apply an attribute classification constraint to the generated image to just guarantee the correct change of desired attributes, i.e., to change what you want. Meanwhile, the reconstruction learning is introduced to preserve attribute-excluding details, in other words, to only change what you want. Besides, the adversarial learning is employed for visually realistic editing. These three components cooperate with each other forming an effective framework for high quality facial attribute editing, referred as AttGAN. Furthermore, the proposed method is extended for attribute style manipulation in an unsupervised manner. Experiments on two wild datasets, CelebA and LFW, show that the proposed method outperforms the state-of-the-art on realistic attribute editing with other facial details well preserved.

Subfields: Computer Vision and Pattern Recognition, Generative Models

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Generative Models

Citations in the corpus

Diffusion Autoencoders: Toward a Meaningful and Decodable Representation (2021, background): “Besides producing impressive image samples, GANs [15] have been shown to learn meaningful latent spaces [27] with extensive studies on multiple derived spaces [24, 57] and various knobs and controls for conditional human face generation [21, 37, 55].”
Generative Counterfactual Introspection for Explainable Deep Learning (2019, methodology): “The classifier [11] is trained on CelebA dataset and achieves average accuracy of 90.”
Editing in Style: Uncovering the Local Semantics of GANs (2020, methodology): “AttGAN [10] uses supervised learning with external attribute classifiers to accomplish attribute editing.”
DeepFake Detection by Analyzing Convolutional Traces (2020, methodology): “In this paper, the ﬁve most famous and effective architectures in state-of-the-art for face Deepfakes were taken into account: STARGAN [6], STYLEGAN [19], STYLE-GAN2 [20], ATTGAN [15] and GDWCT [5].”
Repurposing GANs for One-Shot Semantic Part Segmentation (2021, background): “AttGAN [39] exploits an external classifier network to enable attribute editing.”
Artificial intelligence in the creative industries: a review (2020, background): “2018b), and to change facial attributes, such as the presence of a beard, skin condition, hair style and color (He et al. 2019).”
Multimodal Intelligence: Representation Learning, Information Fusion, and Applications (2019, methodology): “[215] uses the same idea for face editing, such as to remove the beard or change the hair color.”
Image-to-Image Translation: Methods and Applications (2021, background): “Attribute manipulation [159], [170], [167] varies the binary representations to edit image attributes, such as the gender of the subject [224], the color of hair [145], [174] and the presence of glasses [225], and perform image relabeling”
DeepFakes and Beyond: A Survey of Face Manipulation and Fake Detection (2020, background): “Almost at the same time He et al. proposed in [91] attGAN 21 , a novel approach that removes the strict attribute-independent constraint from the latent representation, and just applies the attribute-classiﬁcation constraint to the generated image to guarantee the correct change of the attributes.”
Deepfakes generation and detection: state-of-the-art, open challenges, countermeasures, and way forward (2021, methodology): “In AttGAN [108], an encoder-decoder architecture was proposed that considers the relationship between attributes and latent representation.”
Disentangled Face Attribute Editing via Instance-Aware Latent Space Search (2021, methodology): “, 2017] or network architecture [Choi et al., 2018; Zhang et al., 2018; Liu et al., 2019; Lin et al., 2019; He et al., 2019] manually to improve the output quality.”
A Survey on Generative Adversarial Networks: Variants, Applications, and Training (2020, methodology): “to Editor, a user can make any adjustment on the given photo with the help of rough brush striking. Another high eminence facial characteristic editing framework referred to as Attribute GAN (AttGAN) [73] uses encoder-decoder structural design and takes the facial feature information as a component ”
Only a matter of style: age transformation using a style-based regression model (2021, methodology): “More recent works [He et al. 2018; Lample et al. 2018; Li et al. 2019; Or-El et al. 2020; Yao et al. 2020] take a data-driven approach for modeling face aging by using deep neural networks for image synthesis.”
Implicit Semantic Data Augmentation for Deep Networks (2019, methodology): “Variational Autoencoder(VAE) and Generative Adversarial Network(GAN) based methods [24, 25, 26] establish a latent representation corresponding to the abstractions of images, which can be manipulated to edit the semantics of images.”
Style Intervention: How to Achieve Spatial Disentanglement with Style-based Generators? (2020, methodology): “To solve this problem, numerous methods built on conditional GAN models [8, 12, 17, 18, 28] are proposed, and they typically focus on designing new loss functions or improving network structures.”
Towards Flexible Blind JPEG Artifacts Removal (2021, background): “text-to-image generation [26, 35, 43] and facial attribute editing [9, 21, 27].”
Frequency-Aware Deepfake Detection: Improving Generalizability through Frequency Space Domain Learning (2024, neutral): “AttGAN BEGAN CramerGAN InfoMaxGAN MMDGAN RelGAN S3GAN SNGAN STGAN Mean Acc.”