Shape My Face: Registering 3D Face Scans by Surface-to-Surface Translation

Abstract

Standard registration algorithms need to be independently applied to each surface to register, following careful pre-processing and hand-tuning. Recently, learning-based approaches have emerged that reduce the registration of new scans to running inference with a previously-trained model. The potential benefits are multifold: inference is typically orders of magnitude faster than solving a new instance of a difficult optimization problem, deep learning models can be made robust to noise and corruption, and the trained model may be re-used for other tasks, e.g. through transfer learning. In this paper, we cast the registration task as a surface-to-surface translation problem, and design a model to reliably capture the latent geometric information directly from raw 3D face scans. We introduce Shape-My-Face (SMF), a powerful encoder-decoder architecture based on an improved point cloud encoder, a novel visual attention mechanism, graph convolutional decoders with skip connections, and a specialized mouth model that we smoothly integrate with the mesh convolutions. Compared to the previous state-of-the-art learning algorithms for non-rigid registration of face scans, SMF only requires the raw data to be rigidly aligned (with scaling) with a pre-defined face template. Additionally, our model provides topologically-sound meshes with minimal supervision, offers faster training time, has orders of magnitude fewer trainable parameters, is more robust to noise, and can generalize to previously unseen datasets. We extensively evaluate the quality of our registrations on diverse data. We demonstrate the robustness and generalizability of our model with in-the-wild face scans across different modalities, sensor types, and resolutions. Finally, we show that, by learning to register scans, SMF produces a hybrid linear and non-linear morphable model. Manipulation of the latent space of SMF allows for shape generation, and morphing applications such as expression transfer in-the-wild. We train SMF on a dataset of human faces comprising 9 large-scale databases on commodity hardware.

Static and Motion Facial Analysis for Craniofacial Assessment and Diagnosing Diseases

“…Deep learning-based techniques are in their infancy but have the power to reduce the time complexity considerably. By learning a mapping directly from an input 3D mesh to a standard mesh representation, the expensive, usually iterative, optimization required by more traditional methods is reduced to a single-step inference (28)(29)(30).…”

“…In this way randomly sampled coordinates from the latent space give rise to only realistic shape instances. An alternative approach is to fit a probability model to the distribution of latent-space coordinates, thereby modeling the reasonable limits of the latent-space coordinates on which to base the reconstructions (28,59). Variational autoencoders have also been used to shape the distribution of latent-space coordinates into a user-specified (e.g., multivariate Gaussian) distribution (60).…”

ImFace: A Nonlinear 3D Morphable Face Model with Implicit Neural Representations

“…They indeed deliver performance gains; however, restricted by the resolution of discrete representing strategies on input data, facial priors are not sufficiently captured, incurring loss of shape details. Besides, all current methods are dependent on the preposed procedure of point-to-point correspondence [10], [43], [59], [60], but face registration itself remains challenging.…”

ImFace: A Nonlinear 3D Morphable Face Model with Implicit Neural Representations

“…They indeed deliver performance gains; however, restricted by the resolution of discrete representing strategies on input data, facial priors are not sufficiently captured, incurring loss of shape details. Besides, all current methods are dependent on the preposed procedure of point-to-point correspondence [2,7,26,36], but face registration itself remains challenging.…”