Show simple item record

dc.contributor.authorBoscaini, Davideen_US
dc.contributor.authorMasci, Jonathanen_US
dc.contributor.authorMelzi, Simoneen_US
dc.contributor.authorBronstein, Michael M.en_US
dc.contributor.authorCastellani, Umbertoen_US
dc.contributor.authorVandergheynst, Pierreen_US
dc.contributor.editorMirela Ben-Chen and Ligang Liuen_US
dc.date.accessioned2015-07-06T05:00:06Z
dc.date.available2015-07-06T05:00:06Z
dc.date.issued2015en_US
dc.identifier.urihttp://dx.doi.org/10.1111/cgf.12693en_US
dc.description.abstractIn this paper, we propose a generalization of convolutional neural networks (CNN) to non-Euclidean domains for the analysis of deformable shapes. Our construction is based on localized frequency analysis (a generalization of the windowed Fourier transform to manifolds) that is used to extract the local behavior of some dense intrinsic descriptor, roughly acting as an analogy to patches in images. The resulting local frequency representations are then passed through a bank of filters whose coefficient are determined by a learning procedure minimizing a task-specific cost. Our approach generalizes several previous methods such as HKS, WKS, spectral CNN, and GPS embeddings. Experimental results show that the proposed approach allows learning class-specific shape descriptors significantly outperforming recent state-of-the-art methods on standard benchmarks.en_US
dc.publisherThe Eurographics Association and John Wiley & Sons Ltd.en_US
dc.subjectComputational Geometry and Object Modeling [I.3.5]en_US
dc.subjectFeature Measurement [I.4.7]en_US
dc.subjectLearning [I.2.6]en_US
dc.titleLearning Class-specific Descriptors for Deformable Shapes Using Localized Spectral Convolutional Networksen_US
dc.description.seriesinformationComputer Graphics Forumen_US
dc.description.sectionheadersDescriptors and Shape Synthesisen_US
dc.description.volume34en_US
dc.description.number5en_US
dc.identifier.doi10.1111/cgf.12693en_US
dc.identifier.pages013-023en_US


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record