High Quality Facial Surface and Texture Synthesis via Generative Adversarial Networks

Abstract

In the past several decades, many attempts have been made to model synthetic realistic geometric data. The goal of such models is to generate plausible 3D geometries and textures. Perhaps the best known of its kind is the linear 3D morphable model (3DMM) for faces. Such models can be found at the core of many computer vision applications such as face reconstruction, recognition and authentication to name just a few. Generative adversarial networks (GANs) have shown great promise in imitating high dimensional data distributions. State of the art GANs are capable of performing tasks such as image to image translation as well as auditory and image signal synthesis, producing novel plausible samples from the data distribution at hand. Geometric data is generally more difficult to process due to the inherent lack of an intrinsic parametrization. By bringing geometric data into an aligned space, we are able to map the data onto a 2D plane using a universal parametrization. This alignment process allows for efficient processing of digitally scanned geometric data via image processing tools. Using this methodology, we propose a novel face synthesis model for generation of realistic facial textures together with their corresponding geometry. A GAN is employed in order to imitate the space of parametrized human textures, while corresponding facial geometries are generated by learning the best 3DMM coefficients for each texture. The generated textures are mapped back onto the corresponding geometries to obtain new generated high resolution 3D faces.

Data Augmentation for Leaf Segmentation and Counting Tasks in Rosette Plants

“…For example, Richardson et al based their synthesis on parametric modelling of human faces, constructing synthesized models by learning geometric and texture parameters [8,9,10]. Another paper integrates parametric surface modeling with a Generative Adversarial Network synthesizing realistic facial textures to generate synthetic human faces [11]. Although applied deep learning is common in analysis of plant structure, and computational and heuristic graphical modeling techniques exist, few attempts have been suggested to combine them.…”

Synthesizing Facial Photometries and Corresponding Geometries Using Generative Adversarial Networks

“…Our mapping was also designed in order to take maximal advantage of the available area in each image. In subsection 5.3 we also show that our improved mapping compared to [Slossberg et al 2018] indeed preserves delicate texture details in our predefined high importance regions. In section 6 we also present a new technique for dealing with partially corrupted data.…”

“…The authors of [Slossberg et al 2018] defined tbe mapping between the scan and the plane by using a ray casting technique built into the animation rendering toolbox of Blender [Blender Online Community 2017]. Figure 3 depicts several examples of the resulting mapped facial photometry.…”

“…Lastly, unlike [Slossberg et al 2018], we propose to construct a symmetric mapping, in order to augment the data by mirroring the training samples. This could be done by ensuring that the template is intrinsically symmetric, as well as the choice of boundary vertices and edge weight.…”

“…Next: The deformed template, after its alignment process to some arbitrary scan. Center: the resulting 2D image, using the texture mapping suggested in [Slossberg et al 2018] and the proposed one. Right: A comparison between the texture mapping methods.…”