3D Object Reconstruction from a Single Depth View with Adversarial Learning

TL;DR

This paper introduces 3D-RecGAN, a novel adversarial learning approach that reconstructs complete 3D objects from a single depth view, outperforming existing methods by accurately filling in occluded regions using voxel-based generative models.

Contribution

It presents a new single-view 3D reconstruction method combining autoencoders and GANs, capable of generating detailed 3D structures from minimal input without multiple views.

Findings

Outperforms state-of-the-art in single view 3D reconstruction

Successfully reconstructs unseen object types

Generates high-fidelity 3D occupancy grids

Abstract

In this paper, we propose a novel 3D-RecGAN approach, which reconstructs the complete 3D structure of a given object from a single arbitrary depth view using generative adversarial networks. Unlike the existing work which typically requires multiple views of the same object or class labels to recover the full 3D geometry, the proposed 3D-RecGAN only takes the voxel grid representation of a depth view of the object as input, and is able to generate the complete 3D occupancy grid by filling in the occluded/missing regions. The key idea is to combine the generative capabilities of autoencoders and the conditional Generative Adversarial Networks (GAN) framework, to infer accurate and fine-grained 3D structures of objects in high-dimensional voxel space. Extensive experiments on large synthetic datasets show that the proposed 3D-RecGAN significantly outperforms the state of the art in single view 3D object reconstruction, and is able to reconstruct unseen types of objects. Our code and data are available at: https://github.com/Yang7879/3D-RecGAN.