Improved Adversarial Systems for 3D Object Generation and Reconstruction
arXiv:1707.09557
Abstract
This paper describes a new approach for training generative adversarial networks (GAN) to understand the detailed 3D shape of objects. While GANs have been used in this domain previously, they are notoriously hard to train, especially for the complex joint data distribution over 3D objects of many categories and orientations. Our method extends previous work by employing the Wasserstein distance normalized with gradient penalization as a training objective. This enables improved generation from the joint object shape distribution. Our system can also reconstruct 3D shape from 2D images and perform shape completion from occluded 2.5D range scans. We achieve notable quantitative improvements in comparison to existing baselines
10 pages, accepted at CORL. Figures are best view in color, and details only appear when zoomed in
References in corpus (8)
- Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
- Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks
- Learning a Probabilistic Latent Space of Object Shapes via 3D Generative-Adversarial Modeling
- Improved Training of Wasserstein GANs
- Autoencoding beyond pixels using a learned similarity metric
- Predicting Depth, Surface Normals and Semantic Labels with a Common Multi-Scale Convolutional Architecture
- A Large Dataset of Object Scans
- Unsupervised Learning of 3D Structure from Images
Cited by in corpus (10)
- PCN: Point Completion Network
- Transferring GANs: generating images from limited data
- High Resolution Millimeter Wave Imaging For Self-Driving Cars
- 3D-PhysNet: Learning the Intuitive Physics of Non-Rigid Object Deformations
- Spectral-GANs for High-Resolution 3D Point-cloud Generation
- An Acceleration Framework for High Resolution Image Synthesis
- Paired 3D Model Generation with Conditional Generative Adversarial Networks
- Learning to Generate 3D Shapes with Generative Cellular Automata
- Point Cloud Colorization Based on Densely Annotated 3D Shape Dataset
- Deep Generative Models: Deterministic Prediction with an Application in Inverse Rendering