Modular framework for experimenting with single-view voxel reconstruction, combined with domain adaptation.

A new method to preprocess ShapeNet to get minimal shift 3D ground truth; 3 Stage single-view 3D reconstruction method; Point cloud surface reconstruction without input normals.

Pixel2point in PyTorch

Python implementation of Single View Metrology using numpy and opencv

GAMesh: Guided and Augmented Meshing for Deep Point Networks

[ECCV 2024] SUP-NeRF: A Streamlined Unification of Pose Estimation and NeRF for Monocular 3D Object Reconstruction

Exploring the types of losses and decoder functions for regressing to voxels, point clouds, and mesh representations from single view RGB input.

BuilDiff: 3D Building Shape Generation using Single-Image Conditional Point Cloud Diffusion Models

[AAAI 2023] Official implementation of "Occupancy Planes for Single-view RGB-D Human Reconstruction"

NU-MCC: Multiview Compressive Coding with Neighborhood Decoder and Repulsive UDF

Neural Reflectance Field from Shading and Shadow under a Fixed Viewpoint

Code and datasets for TPAMI 2021 "SkeletonNet: A Topology-Preserving Solution for Learning Mesh Reconstruction of Object Surfaces from RGB Images "

Implementation of CVPR'23: Learning 3D Scene Priors with 2D Supervision

[CVPR 2024] Official implementation of Morphable Diffusion: 3D-Consistent Diffusion for Single-image Avatar Creation

Topologically-Aware Deformation Fields for Single-View 3D Reconstruction (CVPR 2022)

[CVPR 2024] EpiDiff: Enhancing Multi-View Synthesis via Localized Epipolar-Constrained Diffusion

A diffuser implementation of Zero123. Zero-1-to-3: Zero-shot One Image to 3D Object (ICCV23)

(ECCV 2022) Code for Share With Thy Neighbors: Single-View Reconstruction by Cross-Instance Consistency

We evaluate our method on different datasets (including ShapeNet, CUB-200-2011, and Pascal3D+) and achieve state-of-the-art results, outperforming all the other supervised and unsupervised methods and 3D representations, all in terms of performance, accuracy, and training time.

[arXiv'24] VistaDream: Sampling multiview consistent images for single-view scene reconstruction