GAN-Based Face Attribute Editing

Abstract

Recently, a variety of methods using the Generative Adversarial Network (GAN) for face editing have been proposed. However, the existing methods cannot control the editing content of the face elements according to the user-specified attributes or need to train a conditional GAN for editing tasks, which means it is difficult to add new attributes in the future. In this paper, a method to edit face attributes by editing the latent variable with the help of a pre-trained unconditional GAN and a linear classification model is proposed. In particular, face attribute editing is divided into two separate stages: Firstly, based on the optimization function, the generative model does a latent variable search to generate a high-quality face image that is similar to the input image. Secondly, by editing the latent variable of the GAN, the attribute of the generated face image can be modified indirectly, so it is almost unaffected by the training process and network structure of GAN, which means it is a flexible method for nearly all GAN network. Images of the FFHQ dataset are edited by attribute labels defined in Celeba dataset for experiments. These experiments prove that our method can edit a variety of face images that vary with race, gender, age, and camera shooting angle. The overall quality of the edited image is not inferior to the other face attribute editing method, and attribute classification for edited image shows a 92.6% attribute editing success rate of the proposed method.

3D-Guided Face Manipulation of 2D Images for the Prediction of Post-Operative Outcome After Cranio-Maxillofacial Surgery

“…To enable a controlled manipulation of the face, the desired manipulation has to be represented as an interpretable input to the model. To achieve this, Guan [20] and Liu et al [13] both found representations in the input feature vector of GANs to manipulate desired properties of the face image. In contrast, He et al (AttGAN) [14] defined the manipulation by using both discrete attributes as well as a feature vector in latent space to manipulate facial properties of an image.…”

“…8 (a). Additionally, we annotated the first 17 landmarks (# [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17]) to belong to the chin region and nine landmarks (# [28][29][30][31][32][33][34][35][36]) to belong to the nose region as annotated in Fig. 8 (a).…”

“…Hereby, x were the landmark predictions, y were the projected landmarks of the 3D shape, and d = √ w × h was a normalization factor derived by the width w and height h of the bounding boxes given by the AFLW2000 dataset for each unmodified image I n . Additionally, we separately calculated the NME for the chin region using the landmarks # [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17] with N = 17 and for the nose region using the landmarks # [28][29][30][31][32][33][34][35][36] with N = 9 as shown in Fig. 8 (a).…”

“…In contrast to recent state-of-the-art models used to manipulate facial properties, we cannot describe the desired manipulation or a surgery plan by discrete attributes (e.g. brown/blond/black hair color in StarGAN [10], [11]), domain transfer between two images [11], [12], or concealed representations in latent space [13], [14]. Instead, we define the geometric shape of the manipulation using a 3D surface template of the face which enables a precise manipulation according to the 3D shape of the face.…”

Attribute-Aware Generative Design With Generative Adversarial Networks

“…Attribute editing with GAN involves making translations/adjustments to images based on the target attributes to generate a new sample with desired attributes while preserving other details of the original image. Current GAN-based attribute editing research is predominantly centered on human face images [18]- [20]. The facial attribute editing task allows to edit a face image by manipulating one or multiple attributes of interest such as hair color, expression, mustache, and age [21], [22].…”