A Convolutional Decoder for Point Clouds using Adaptive Instance Normalization

dc.contributor.authorLim, Isaaken_US
dc.contributor.authorIbing, Moritzen_US
dc.contributor.authorKobbelt, Leifen_US
dc.contributor.editorBommes, David and Huang, Huien_US
dc.date.accessioned2019-07-11T06:19:17Z
dc.date.available2019-07-11T06:19:17Z
dc.date.issued2019
dc.description.abstractAutomatic synthesis of high quality 3D shapes is an ongoing and challenging area of research. While several data-driven methods have been proposed that make use of neural networks to generate 3D shapes, none of them reach the level of quality that deep learning synthesis approaches for images provide. In this work we present a method for a convolutional point cloud decoder/generator that makes use of recent advances in the domain of image synthesis. Namely, we use Adaptive Instance Normalization and offer an intuition on why it can improve training. Furthermore, we propose extensions to the minimization of the commonly used Chamfer distance for auto-encoding point clouds. In addition, we show that careful sampling is important both for the input geometry and in our point cloud generation process to improve results. The results are evaluated in an autoencoding setup to offer both qualitative and quantitative analysis. The proposed decoder is validated by an extensive ablation study and is able to outperform current state of the art results in a number of experiments. We show the applicability of our method in the fields of point cloud upsampling, single view reconstruction, and shape synthesis.en_US
dc.description.number5
dc.description.sectionheadersSynthesis and Learning
dc.description.seriesinformationComputer Graphics Forum
dc.description.volume38
dc.identifier.doi10.1111/cgf.13792
dc.identifier.issn1467-8659
dc.identifier.pages99-108
dc.identifier.urihttps://doi.org/10.1111/cgf.13792
dc.identifier.urihttps://diglib.eg.org:443/handle/10.1111/cgf13792
dc.publisherThe Eurographics Association and John Wiley & Sons Ltd.en_US
dc.subjectComputing methodologies
dc.subjectShape analysis
dc.subjectPoint
dc.subjectbased models
dc.titleA Convolutional Decoder for Point Clouds using Adaptive Instance Normalizationen_US
Files
Collections