Abstract:A signed distance function (SDF) as the 3D shape description is one of the most effective approaches to represent 3D geometry for rendering and reconstruction. Our work is inspired by the state-of-the-art method DeepSDF [17] that learns and analyzes the 3D shape as the iso-surface of its shell and this method has shown promising results especially in the 3D shape reconstruction and compression domain. In this paper, we consider the degeneration problem of reconstruction coming from the capacity decrease of the DeepSDF model, which approximates the SDF with a neural network and a single latent code. We propose Local Geometry Code Learning (LGCL), a model that improves the original DeepSDF results by learning from a local shape geometry of the full 3D shape. We add an extra graph neural network to split the single transmittable latent code into a set of local latent codes distributed on the 3D shape. Mentioned latent codes are used to approximate the SDF in their local regions, which will alleviate the complexity of the approximation compared to the original DeepSDF. Furthermore, we introduce a new geometric loss function to facilitate the training of these local latent codes. Note that other local shape adjusting methods use the 3D voxel representation, which in turn is a problem highly difficult to solve or even is insolvable. In contrast, our architecture is based on graph processing implicitly and performs the learning regression process directly in the latent code space, thus make the proposed architecture more flexible and also simple for realization. Our experiments on 3D shape reconstruction demonstrate that our LGCL method can keep more details with a significantly smaller size of the SDF decoder and outperforms considerably the original DeepSDF method under the most important quantitative metrics.

Deep Marching Cubes: Learning Explicit Surface Representations.

Deep Local Shapes: Learning Local SDF Priors for Detailed 3D Reconstruction

Learning to Infer Implicit Surfaces without 3D Supervision

DiffusionNet: Discretization Agnostic Learning on Surfaces

Hierarchical Surface Prediction for 3D Object Reconstruction

DeepSDF: Learning Continuous Signed Distance Functions for Shape Representation

Deep Learning 3d Shape Surfaces Using Geometry Images

Learning Deformable Tetrahedral Meshes for 3D Reconstruction.

SurfNet: Generating 3D shape surfaces using deep residual networks

Deep Learning 3D Shapes Using Alt-az Anisotropic 2-Sphere Convolution

3D ShapeNets: A Deep Representation for Volumetric Shapes

Shape As Points: A Differentiable Poisson Solver

NeAT: Learning Neural Implicit Surfaces with Arbitrary Topologies from Multi-view Images

Quadratic Terms Based Point-to-Surface 3D Representation for Deep Learning of Point Cloud

Voxel-Based 3D Shape Segmentation Using Deep Volumetric Convolutional Neural Networks

3D Shapes Local Geometry Codes Learning with SDF

Learning to Infer and Execute 3D Shape Programs

LISR: Learning Linear 3D Implicit Surface Representation Using Compactly Supported Radial Basis Functions

D eep l earning 3 d s hapes u sing a lt-az a nisotropic 2s phere c onvolution

Deep Implicit Moving Least-Squares Functions for 3D Reconstruction

Patch-Grid: An Efficient and Feature-Preserving Neural Implicit Surface Representation