3D VAE-Attention Network: A Parallel System for Single-view 3D Reconstruction

Simple item page
dc.contributor.authorHu, Feien_US
dc.contributor.authorYang, Xinyanen_US
dc.contributor.authorZhong, Weien_US
dc.contributor.authorYe, Longen_US
dc.contributor.authorZhang, Qinen_US
dc.contributor.editorFu, Hongbo and Ghosh, Abhijeet and Kopf, Johannesen_US
dc.date.accessioned2018-10-07T14:32:11Z
dc.date.available2018-10-07T14:32:11Z
dc.date.issued2018
dc.description.abstract3D object reconstruction from single view image is a challenge task. Due to the fact that the information contained in one isolated image is not sufficient for reasonable 3D shape reconstruction, the existing results on single-view 3D reconstruction always lack marginal voxels. To tackle this problem, we propose a parallel system named 3D VAE-attention network (3VAN) for single view 3D reconstruction. Distinct from the common encoder-decoder structure, the proposed network consists of two parallel branches, 3D-VAE and Attention Network. 3D-VAE completes the general shape reconstruction by an extension of standard VAE model, and Attention Network supplements the missing details by a 3D reconstruction attention network. In the experiments, we verify the feasibility of our 3VAN on the ShapeNet and PASCAL 3D+ datasets. By comparing with the state-of-art methods, the proposed 3VAN can produce more precise 3D object models in terms of both qualitative and quantitative evaluation.en_US
dc.description.sectionheaders3D Modeling
dc.description.seriesinformationPacific Graphics Short Papers
dc.identifier.doi10.2312/pg.20181279
dc.identifier.isbn978-3-03868-073-4
dc.identifier.pages53-56
dc.identifier.urihttps://doi.org/10.2312/pg.20181279
dc.identifier.urihttps://diglib.eg.org:443/handle/10.2312/pg20181279
dc.publisherThe Eurographics Associationen_US
dc.subjectComputing methodologies
dc.subjectReconstruction
dc.subjectVolumetric models
dc.title3D VAE-Attention Network: A Parallel System for Single-view 3D Reconstructionen_US
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
053-056.pdf
Size:
24.14 MB
Format:
Adobe Portable Document Format
Download
Collections
PG2018 Short Papers and Posters