Learning an interpretable factorised representation of the independent data generative factors of the world without supervision is an important precursor for the development of artificial intelligence that is able to learn and reason in the same way that humans do. We introduce beta-VAE, a new state-of-the-art framework for automated discovery of interpretable factorised latent representations from raw image data in a completely unsupervised manner. Our approach is a modification of the variational autoencoder (VAE) framework. We introduce an adjustable hyperparameter beta that balances latent channel capacity and independence constraints with reconstruction accuracy. We demonstrate that beta-VAE with appropriately tuned beta > 1 qualitatively outperforms VAE (beta = 1), as well as state of the art unsupervised (InfoGAN) and semi-supervised (DC-IGN) approaches to disentangled factor learning on a variety of datasets (celebA, faces and chairs). Furthermore, we devise a protocol to quantitatively compare the degree of disentanglement learnt by different models, and show that our approach also significantly outperforms all baselines quantitatively. Unlike InfoGAN, beta-VAE is stable to train, makes few assumptions about the data and relies on tuning a single hyperparameter, which can be directly optimised through a hyper parameter search using weakly labelled data or through heuristic visual inspection for purely unsupervised data.

beta-VAE: Learning Basic Visual Concepts with a Constrained Variational Framework

This paper describes InfoGAN, an information-theoretic extension to the Generative Adversarial Network that is able to learn disentangled representations in a completely unsupervised manner. InfoGAN is a generative adversarial network that also maximizes the mutual information between a small subset of the latent variables and the observation. We derive a lower bound to the mutual information objective that can be optimized efficiently, and show that our training procedure can be interpreted as a variation of the Wake-Sleep algorithm. Specifically, InfoGAN successfully disentangles writing styles from digit shapes on the MNIST dataset, pose from lighting of 3D rendered images, and background digits from the central digit on the SVHN dataset. It also discovers visual concepts that include hair styles, presence/absence of eyeglasses, and emotions on the CelebA face dataset. Experiments show that InfoGAN learns interpretable representations that are competitive with representations learned by existing fully supervised methods.

InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets

This paper describes InfoGAN, an information-theoretic extension to the Generative Adversarial Network that is able to learn disentangled representations in a completely unsupervised manner. InfoGAN is a generative adversarial network that also maximizes the mutual information between a small subset of the latent variables and the observation. We derive a lower bound of the mutual information objective that can be optimized efficiently. Specifically, InfoGAN successfully disentangles writing styles from digit shapes on the MNIST dataset, pose from lighting of 3D rendered images, and background digits from the central digit on the SVHN dataset. It also discovers visual concepts that include hair styles, presence/absence of eyeglasses, and emotions on the CelebA face dataset. Experiments show that InfoGAN learns interpretable representations that are competitive with representations learned by existing supervised methods. For an up-to-date version of this paper, please see https://arxiv.org/abs/1606.03657.

/pdf/infogan-interpretable-representation-learning-by-information-5apd70hss0.pdf

InfoGAN: interpretable representation learning by information maximizing generative adversarial nets

IEEE Transactions on Pattern Analysis and Machine Intelligence

A morphable model for the synthesis of 3D faces

Generative 3D face models are a powerful tool in computer vision. They provide pose and illumination invariance by modeling the space of 3D faces and the imaging process. The power of these models comes at the cost of an expensive and tedious construction process, which has led the community to focus on more easily constructed but less powerful models. With this paper we publish a generative 3D shape and texture model, the Basel Face Model (BFM), and demonstrate its application to several face recognition task. We improve on previous models by offering higher shape and texture accuracy due to a better scanning device and less correspondence artifacts due to an improved registration algorithm. The same 3D face model can be fit to 2D or 3D images acquired under different situations and with different sensors using an analysis by synthesis method. The resulting model parameters separate pose, lighting, imaging and identity parameters, which facilitates invariant face recognition across sensors and data sets by comparing only the identity parameters. We hope that the availability of this registered face model will spur research in generative models. Together with the model we publish a set of detailed recognition and reconstruction results on standard databases to allow complete algorithm comparisons.

A 3D Face Model for Pose and Illumination Invariant Face Recognition

We show how to extend the ICP framework to nonrigid registration, while retaining the convergence properties of the original algorithm. The resulting optimal step nonrigid ICP framework allows the use of different regularisations, as long as they have an adjustable stiffness parameter. The registration loops over a series of decreasing stiffness weights, and incrementally deforms the template towards the target, recovering the whole range of global and local deformations. To find the optimal deformation for a given stiffness, optimal iterative closest point steps are used. Preliminary correspondences are estimated by a nearest-point search. Then the optimal deformation of the template for these fixed correspondences and the active stiffness is calculated. Afterwards the process continues with new correspondences found by searching from the displaced template vertices. We present an algorithm using a locally affine regularisation which assigns an affine transformation to each vertex and minimises the difference in the transformation of neighbouring vertices. It is shown that for this regularisation the optimal deformation for fixed correspondences and fixed stiffness can be determined exactly and efficiently. The method succeeds for a wide range of initial conditions, and handles missing data robustly. It is compared qualitatively and quantitatively to other algorithms using synthetic examples and real world data.

/pdf/optimal-step-nonrigid-icp-algorithms-for-surface-zl320phu5d.pdf

Optimal Step Nonrigid ICP Algorithms for Surface Registration

We present a novel algorithm aiming to estimate the 3D shape, the texture of a human face, along with the 3D pose and the light direction from a single photograph by recovering the parameters of a 3D morphable model. Generally, the algorithms tackling the problem of 3D shape estimation from image data use only the pixels intensity as input to drive the estimation process. This was previously achieved using either a simple model, such as the Lambertian reflectance model, leading to a linear fitting algorithm. Alternatively, this problem was addressed using a more precise model and minimizing a non-convex cost function with many local minima. One way to reduce the local minima problem is to use a stochastic optimization algorithm. However, the convergence properties (such as the radius of convergence) of such algorithms, are limited. Here, as well as the pixel intensity, we use various image features such as the edges or the location of the specular highlights. The 3D shape, texture and imaging parameters are then estimated by maximizing the posterior of the parameters given these image features. The overall cost function obtained is smoother and, hence, a stochastic optimization algorithm is not needed to avoid the local minima problem. This leads to the multi-features fitting algorithm that has a wider radius of convergence and a higher level of precision. This is shown on some example photographs, and on a recognition experiment performed on the CMU-PIE image database.

Estimating 3D shape and texture using pixel intensity, edges, specular highlights, texture constraints and a prior

This paper describes an algorithm for finding faces within an image. The basis of the algorithm is to run an observation window at all possible positions, scales and orientation within the image. A non-linear support vector machine is used to determine whether or not a face is contained within the observation window. The non-linear support vector machine operates by comparing the input patch to a set of support vectors (which can be thought of as face and anti-face templates). Each support vector is scored by some nonlinear function against the observation window and if the resulting sum is over some threshold a face is indicated. Because of the huge search space that is considered, it is imperative to investigate ways to speed up the support vector machine. Within this paper we suggest a method of speeding up the non-linear support vector machine. A set of reduced set vectors (RVs) are calculated from the support vectors. By considering the RV's sequentially, and if at any point a face is deemed too unlikely to cease the sequential evaluation, obviating the need to evaluate the remaining RVs. The idea being that we only need to apply a subset of the RVs to eliminate things that are obviously not a face (thus reducing the computation). The key then is to explore the RVs in the right order and a method for this is proposed.

/pdf/computationally-efficient-face-detection-37pl7j1c30.pdf

Computationally efficient face detection

In this article, we provide a detailed survey of 3D Morphable Face Models over the 20 years since they were first proposed. The challenges in building and applying these models, namely, capture, modeling, image formation, and image analysis, are still active research topics, and we review the state-of-the-art in each of these areas. We also look ahead, identifying unsolved challenges, proposing directions for future research, and highlighting the broad range of current and future applications.

/pdf/3d-morphable-face-models-past-present-and-future-33bjgatyz8.pdf

Sami Romdhani

Papers

A 3D Face Model for Pose and Illumination Invariant Face Recognition

Optimal Step Nonrigid ICP Algorithms for Surface Registration

Estimating 3D shape and texture using pixel intensity, edges, specular highlights, texture constraints and a prior

Computationally efficient face detection

3D Morphable Face Models—Past, Present, and Future