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

Local Deep Feature Learning Framework for 3D Shape.

Local Deep Implicit Functions for 3D Shape

3D Shapes Local Geometry Codes Learning with SDF

DeepSDF: Learning Continuous Signed Distance Functions for Shape Representation

Deep Optimized Priors for 3D Shape Modeling and Reconstruction

ResDepth: A Deep Residual Prior For 3D Reconstruction From High-resolution Satellite Images

SDF-SRN: Learning Signed Distance 3D Object Reconstruction from Static Images

3D Surface Reconstruction in the Wild by Deforming Shape Priors from Synthetic Data

Deep-Learning-Based 3-D Surface Reconstruction—A Survey

DeepShape: Deep-Learned Shape Descriptor for 3D Shape Retrieval

Curriculum DeepSDF

DASI: Learning Domain Adaptive Shape Impression for 3D Object Reconstruction

DISN: Deep Implicit Surface Network for High-quality Single-view 3D Reconstruction

Deep Learned Full-3D Object Completion from Single View

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

Deep Learning 3d Shape Surfaces Using Geometry Images

Learning Free-Form Deformations for 3D Object Reconstruction

FineRecon: Depth-aware Feed-forward Network for Detailed 3D Reconstruction

SurfNet: Generating 3D shape surfaces using deep residual networks

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