3D-CODED : 3D Correspondences by Deep Deformation

Abstract : We present a new deep learning approach for matching deformable shapes by introducing Shape Deformation Networks which jointly encode 3D shapes and correspondences. This is achieved by factoring the surface representation into (i) a template, that parameterizes the surface, and (ii) a learnt global feature vector that parameterizes the transformation of the template into the input surface. By predicting this feature for a new shape, we implicitly predict correspondences between this shape and the template. We show that these correspondences can be improved by an additional step which improves the shape feature by minimizing the Chamfer distance between the input and transformed template. We demonstrate that our simple approach improves on state-of-the-art results on the difficult FAUST-inter challenge, with an average correspondence error of 2.88cm. We show, on the TOSCA dataset, that our method is robust to many types of perturbations, and generalizes to non-human shapes. This robustness allows it to perform well on real unclean, meshes from the the SCAPE dataset.
Complete list of metadatas

Cited literature [31 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-01830474
Contributor : Thibault Groueix <>
Submitted on : Tuesday, August 28, 2018 - 2:07:05 PM
Last modification on : Tuesday, March 5, 2019 - 2:40:07 PM
Long-term archiving on : Thursday, November 29, 2018 - 4:02:59 PM

File

1806.05228 (1).pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01830474, version 2
  • ARXIV : 1806.05228

Collections

Citation

Thibault Groueix, Matthew Fisher, Vladimir Kim, Bryan Russell, Mathieu Aubry. 3D-CODED : 3D Correspondences by Deep Deformation. ECCV 2018, Sep 2018, Munich, Germany. ⟨hal-01830474v2⟩

Share

Metrics

Record views

134

Files downloads

192