Lifting AutoEncoders

Unsupervised Learning of a Fully-Disentangled 3D Morphable Model
using Deep Non-Rigid Structure from Motion
arXiv preprint

  • signifies equal contribution

Abstract

In this work we introduce Lifting AutoEncoders, a generative 3D surface-based model of object categories. We bring together ideas from non-rigid structure from motion, image formation, and morphable models to learn a controllable, geometric model of 3D categories in an entirely unsupervised manner from an unstructured set of images. We exploit the 3D geometric nature of our model and use normal information to disentangle appearance into illumination, shading and albedo. We further use weak supervision to disentangle the non-rigid shape variability of human faces into identity and expression. We combine the 3D representation with a differentiable renderer to generate RGB images and append an adversarially trained refinement network to obtain sharp, photorealistic image reconstruction results. The learned generative model can be controlled in terms of interpretable geometry and appearance factors, allowing us to perform photorealistic image manipulation of identity, expression, 3D pose, and illumination properties.



Shape from Unconstrained Images

p1



Network Architectures

Shape learning using reprojection loss

p1

Geometry-based disentanglement of albedo and shading

p1



Training on CelebA
fully unsupervised

Interpolation over texture, pose, and shape

p1

Visualisations from different viewpoints

p1

Lighting manipulation

p1



Training on MultiPIE
weak supervision for identity, pose, and expression

Expression manipulation

p1

Pose manipulation

p1

Downloads