GLASS: Geometric Latent Augmentation for Shape Spaces

TL;DR

This paper introduces GLASS, a method that uses geometric energies to augment sparse 3D shape datasets, enabling effective training of generative models like VAEs for shape synthesis and correspondence with very limited data.

Contribution

GLASS proposes a novel geometric augmentation framework that enhances sparse 3D shape collections, improving generative modeling and shape analysis from minimal data.

Findings

Effective shape generation from as few as 3-10 models.

Outperforms strong baselines in shape synthesis tasks.

Enables shape correspondence with limited data.

Abstract

We investigate the problem of training generative models on a very sparse collection of 3D models. We use geometrically motivated energies to augment and thus boost a sparse collection of example (training) models. We analyze the Hessian of the as-rigid-as-possible (ARAP) energy to sample from and project to the underlying (local) shape space, and use the augmented dataset to train a variational autoencoder (VAE). We iterate the process of building latent spaces of VAE and augmenting the associated dataset, to progressively reveal a richer and more expressive generative space for creating geometrically and semantically valid samples. Our framework allows us to train generative 3D models even with a small set of good quality 3D models, which are typically hard to curate. We extensively evaluate our method against a set of strong baselines, provide ablation studies and demonstrate application towards establishing shape correspondences. We present multiple examples of interesting and meaningful shape variations even when starting from as few as 3-10 training shapes.