Learning Gradient Fields for Shape Generation
arXiv:2008.06520
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. Code is available at https://github.com/RuojinCai/ShapeGF.
Published in ECCV 2020 (Spotlight); Project page: https://www.cs.cornell.edu/~ruojin/ShapeGF/
References in corpus (10)
- Conditional Image Generation with PixelCNN Decoders
- Pixel Recurrent Neural Networks
- Learning a Probabilistic Latent Space of Object Shapes via 3D Generative-Adversarial Modeling
- Variational Inference with Normalizing Flows
- MeshCNN: A Network with an Edge
- Learning Representations and Generative Models for 3D Point Clouds
- Stochastic Gradient Hamiltonian Monte Carlo
- Normalizing Flows for Probabilistic Modeling and Inference
- Implicit Generation and Generalization in Energy-Based Models
- Deep Level Sets: Implicit Surface Representations for 3D Shape Inference