Learning Gradient Fields for Shape Generation
Abstract
In this work, we propose a novel technique to generate shapes from point cloud data. A point cloud can be viewed as samples from a distribution of 3D points whose density is concentrated near the surface of the shape. Point cloud generation thus amounts to moving randomly sampled points to high-density areas. We generate point clouds by performing stochastic gradient ascent on an unnormalized probability density, thereby moving sampled points toward the high-likelihood regions. Our model directly predicts the gradient of the log density field and can be trained with a simple objective adapted from score-based generative models. We show that our method can reach state-of-the-art performance for point cloud auto-encoding and generation, while also allowing for extraction of a high-quality implicit surface.
Cite
Text
Cai et al. "Learning Gradient Fields for Shape Generation." Proceedings of the European Conference on Computer Vision (ECCV), 2020. doi:10.1007/978-3-030-58580-8_22Markdown
[Cai et al. "Learning Gradient Fields for Shape Generation." Proceedings of the European Conference on Computer Vision (ECCV), 2020.](https://mlanthology.org/eccv/2020/cai2020eccv-learning-a/) doi:10.1007/978-3-030-58580-8_22BibTeX
@inproceedings{cai2020eccv-learning-a,
title = {{Learning Gradient Fields for Shape Generation}},
author = {Cai, Ruojin and Yang, Guandao and Averbuch-Elor, Hadar and Hao, Zekun and Belongie, Serge and Snavely, Noah and Hariharan, Bharath},
booktitle = {Proceedings of the European Conference on Computer Vision (ECCV)},
year = {2020},
doi = {10.1007/978-3-030-58580-8_22},
url = {https://mlanthology.org/eccv/2020/cai2020eccv-learning-a/}
}