Learning Gradient Fields for Shape Generation
Research output: Contribution to journal › Conference article › Research › peer-review
Standard
Learning Gradient Fields for Shape Generation. / Cai, Ruojin; Yang, Guandao; Averbuch-Elor, Hadar; Hao, Zekun; Belongie, Serge; Snavely, Noah; Hariharan, Bharath.
In: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2020, p. 364-381.Research output: Contribution to journal › Conference article › Research › peer-review
Harvard
APA
Vancouver
Author
Bibtex
}
RIS
TY - GEN
T1 - Learning Gradient Fields for Shape Generation
AU - Cai, Ruojin
AU - Yang, Guandao
AU - Averbuch-Elor, Hadar
AU - Hao, Zekun
AU - Belongie, Serge
AU - Snavely, Noah
AU - Hariharan, Bharath
N1 - Funding Information: Acknowledgment. This work was supported in part by grants from Magic Leap and Facebook AI, and the Zuckerman STEM leadership program. Publisher Copyright: © 2020, Springer Nature Switzerland AG.
PY - 2020
Y1 - 2020
N2 - 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. Code is available at https://github.com/RuojinCai/ShapeGF.
AB - 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. Code is available at https://github.com/RuojinCai/ShapeGF.
KW - 3D generation
KW - Generative models
UR - http://www.scopus.com/inward/record.url?scp=85097830479&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-58580-8_22
DO - 10.1007/978-3-030-58580-8_22
M3 - Conference article
AN - SCOPUS:85097830479
SP - 364
EP - 381
JO - Lecture Notes in Computer Science
JF - Lecture Notes in Computer Science
SN - 0302-9743
T2 - 16th European Conference on Computer Vision, ECCV 2020
Y2 - 23 August 2020 through 28 August 2020
ER -
ID: 301818367