3D object reconstruction from a single depth view with adversarial learning
<p>In this paper, we propose a novel <b>3D-RecGAN</b> approach, which reconstructs the complete 3D structure of a given object from a single arbitrary depth view using generative adversarial networks. Unlike the existing work which typically requires multiple views of the same obj...
Main Authors: | , , , , , |
---|---|
Format: | Conference item |
Published: |
IEEE
2018
|
_version_ | 1797089604889739264 |
---|---|
author | Yang, B Wen, H Wang, S Clark, R Markham, A Trigoni, N |
author_facet | Yang, B Wen, H Wang, S Clark, R Markham, A Trigoni, N |
author_sort | Yang, B |
collection | OXFORD |
description | <p>In this paper, we propose a novel <b>3D-RecGAN</b> approach, which reconstructs the complete 3D structure of a given object from a single arbitrary depth view using generative adversarial networks. Unlike the existing work which typically requires multiple views of the same object or class labels to recover the full 3D geometry, the proposed 3D-RecGAN only takes the voxel grid representation of a depth view of the object as input, and is able to generate the complete 3D occupancy grid by filling in the occluded/missing regions. The key idea is to combine the generative capabilities of autoencoders and the conditional Generative Adversarial Networks (GAN) framework, to infer accurate and finegrained 3D structures of objects in high-dimensional voxel space. Extensive experiments on large synthetic datasets show that the proposed 3D-RecGAN significantly outperforms the state of the art in single view 3D object reconstruction, and is able to reconstruct unseen types of objects. Our code and data are available at: https://github. com/Yang7879/3D-RecGAN</p> |
first_indexed | 2024-03-07T03:06:26Z |
format | Conference item |
id | oxford-uuid:b2b17a79-ded2-4a64-bc5f-db5a75736bbd |
institution | University of Oxford |
last_indexed | 2024-03-07T03:06:26Z |
publishDate | 2018 |
publisher | IEEE |
record_format | dspace |
spelling | oxford-uuid:b2b17a79-ded2-4a64-bc5f-db5a75736bbd2022-03-27T04:13:31Z3D object reconstruction from a single depth view with adversarial learningConference itemhttp://purl.org/coar/resource_type/c_5794uuid:b2b17a79-ded2-4a64-bc5f-db5a75736bbdSymplectic Elements at OxfordIEEE2018Yang, BWen, HWang, SClark, RMarkham, ATrigoni, N <p>In this paper, we propose a novel <b>3D-RecGAN</b> approach, which reconstructs the complete 3D structure of a given object from a single arbitrary depth view using generative adversarial networks. Unlike the existing work which typically requires multiple views of the same object or class labels to recover the full 3D geometry, the proposed 3D-RecGAN only takes the voxel grid representation of a depth view of the object as input, and is able to generate the complete 3D occupancy grid by filling in the occluded/missing regions. The key idea is to combine the generative capabilities of autoencoders and the conditional Generative Adversarial Networks (GAN) framework, to infer accurate and finegrained 3D structures of objects in high-dimensional voxel space. Extensive experiments on large synthetic datasets show that the proposed 3D-RecGAN significantly outperforms the state of the art in single view 3D object reconstruction, and is able to reconstruct unseen types of objects. Our code and data are available at: https://github. com/Yang7879/3D-RecGAN</p> |
spellingShingle | Yang, B Wen, H Wang, S Clark, R Markham, A Trigoni, N 3D object reconstruction from a single depth view with adversarial learning |
title | 3D object reconstruction from a single depth view with adversarial learning |
title_full | 3D object reconstruction from a single depth view with adversarial learning |
title_fullStr | 3D object reconstruction from a single depth view with adversarial learning |
title_full_unstemmed | 3D object reconstruction from a single depth view with adversarial learning |
title_short | 3D object reconstruction from a single depth view with adversarial learning |
title_sort | 3d object reconstruction from a single depth view with adversarial learning |
work_keys_str_mv | AT yangb 3dobjectreconstructionfromasingledepthviewwithadversariallearning AT wenh 3dobjectreconstructionfromasingledepthviewwithadversariallearning AT wangs 3dobjectreconstructionfromasingledepthviewwithadversariallearning AT clarkr 3dobjectreconstructionfromasingledepthviewwithadversariallearning AT markhama 3dobjectreconstructionfromasingledepthviewwithadversariallearning AT trigonin 3dobjectreconstructionfromasingledepthviewwithadversariallearning |