Improved Adversarial Systems for 3D Object Generation and Reconstruction

Edward J. Smith, David Meger
Proceedings of the 1st Annual Conference on Robot Learning, PMLR 78:87-96, 2017.

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.

Cite this Paper

BibTeX
@InProceedings{pmlr-v78-smith17a, title = {Improved Adversarial Systems for 3D Object Generation and Reconstruction}, author = {Smith, Edward J. and Meger, David}, booktitle = {Proceedings of the 1st Annual Conference on Robot Learning}, pages = {87--96}, year = {2017}, editor = {Levine, Sergey and Vanhoucke, Vincent and Goldberg, Ken}, volume = {78}, series = {Proceedings of Machine Learning Research}, month = {13--15 Nov}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v78/smith17a/smith17a.pdf}, url = {https://proceedings.mlr.press/v78/smith17a.html}, 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.} }
Endnote
%0 Conference Paper %T Improved Adversarial Systems for 3D Object Generation and Reconstruction %A Edward J. Smith %A David Meger %B Proceedings of the 1st Annual Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2017 %E Sergey Levine %E Vincent Vanhoucke %E Ken Goldberg %F pmlr-v78-smith17a %I PMLR %P 87--96 %U https://proceedings.mlr.press/v78/smith17a.html %V 78 %X 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.
APA
Smith, E.J. & Meger, D.. (2017). Improved Adversarial Systems for 3D Object Generation and Reconstruction. Proceedings of the 1st Annual Conference on Robot Learning, in Proceedings of Machine Learning Research 78:87-96 Available from https://proceedings.mlr.press/v78/smith17a.html.

Related Material