Monocular 3D Object Reconstruction with GAN Inversion

TL;DR

MeshInversion leverages a pre-trained 3D GAN to reconstruct textured 3D meshes from single images by searching the GAN's latent space, improving fidelity and generalization for in-the-wild objects.

Contribution

This work introduces MeshInversion, a novel framework that uses GAN inversion in 3D space for monocular textured mesh reconstruction, enhancing realism and handling unseen object variations.

Findings

Achieves faithful 3D reconstructions with consistent geometry and texture.

Generalizes well to less common and deformable object meshes.

Outperforms existing methods on standard benchmarks.

Abstract

Recovering a textured 3D mesh from a monocular image is highly challenging, particularly for in-the-wild objects that lack 3D ground truths. In this work, we present MeshInversion, a novel framework to improve the reconstruction by exploiting the generative prior of a 3D GAN pre-trained for 3D textured mesh synthesis. Reconstruction is achieved by searching for a latent space in the 3D GAN that best resembles the target mesh in accordance with the single view observation. Since the pre-trained GAN encapsulates rich 3D semantics in terms of mesh geometry and texture, searching within the GAN manifold thus naturally regularizes the realness and fidelity of the reconstruction. Importantly, such regularization is directly applied in the 3D space, providing crucial guidance of mesh parts that are unobserved in the 2D space. Experiments on standard benchmarks show that our framework obtains faithful 3D reconstructions with consistent geometry and texture across both observed and unobserved parts. Moreover, it generalizes well to meshes that are less commonly seen, such as the extended articulation of deformable objects. Code is released at https://github.com/junzhezhang/mesh-inversion