Unsupervised Learning of 3D Structure from Images

TL;DR

This paper introduces a novel unsupervised deep learning approach to infer 3D structures from 2D images, achieving high-quality results and establishing new benchmarks in the field.

Contribution

It presents the first end-to-end trainable deep generative models for 3D structure inference from images without supervision.

Findings

High-quality 3D samples generated

Achieved competitive log-likelihoods on ShapeNet

First benchmarks for unsupervised 3D inference

Abstract

A key goal of computer vision is to recover the underlying 3D structure from 2D observations of the world. In this paper we learn strong deep generative models of 3D structures, and recover these structures from 3D and 2D images via probabilistic inference. We demonstrate high-quality samples and report log-likelihoods on several datasets, including ShapeNet [2], and establish the first benchmarks in the literature. We also show how these models and their inference networks can be trained end-to-end from 2D images. This demonstrates for the first time the feasibility of learning to infer 3D representations of the world in a purely unsupervised manner.