luethan2025/3dgan

Quick Start

git clone https://github.com/bareform/3dgan.git
cd 3dgan

You can download the pre-trained models here and use the provided Jupyter Notebook inference.ipynb to generate some samples.

Method

Learning a Probabilistic Latent Space of Object Shapes via 3D Generative-Adversarial Modeling

Jiajun Wu¹, Chengkai Zhang¹, Tianfan Xue¹, William T. Freeman¹², Joshua B. Tenenbaum¹

¹MIT Computer Science & Artificial Intelligence Laboratory, ²Google Research

We study the problem of 3D object generation. We propose a novel framework, namely 3D Generative Adversarial Network (3D-GAN), which generates 3D objects from a probabilistic space by leveraging recent advances in volumetric convolutional networks and generative adversarial nets. The benefits of our model are three-fold: first, the use of an adversarial criterion, instead of traditional heuristic criteria, enables the generator to capture object structure implicitly and to synthesize high-quality 3D objects; second, the generator establishes a mapping from a low-dimensional probabilistic space to the space of 3D objects, so that we can sample objects without a reference image or CAD models, and explore the 3D object manifold; third, the adversarial discriminator provides a powerful 3D shape descriptor which, learned without supervision, has wide applications in 3D object recognition. Experiments demonstrate that our method generates high-quality 3D objects, and our unsupervisedly learned features achieve impressive performance on 3D object recognition, comparable with those of supervised learning methods.

Citation

The original paper can be found at:

@misc{wu20163dgan,
    title={Learning a Probabilistic Latent Space of Object Shapes via 3D Generative-Adversarial Modeling},
    author={Jiajun Wu and Chengkai Zhang and Tianfan Xue and William T. Freeman and Joshua B. Tenenbaum},
    year={2016},
    eprint={1610.07584},
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}