FeaStNet: Feature-Steered Graph Convolutions for 3D Shape Analysis

Nitika Verma 1, 2 Edmond Boyer 1 Jakob Verbeek 2
1 MORPHEO - Capture and Analysis of Shapes in Motion
Inria Grenoble - Rhône-Alpes, LJK - Laboratoire Jean Kuntzmann, INPG - Institut National Polytechnique de Grenoble
2 Thoth - Apprentissage de modèles à partir de données massives
Inria Grenoble - Rhône-Alpes, LJK - Laboratoire Jean Kuntzmann
Abstract : Convolutional neural networks (CNNs) have massively impacted visual recognition in 2D images, and are now ubiquitous in state-of-the-art approaches. CNNs do not easily extend, however, to data that are not represented by regular grids, such as 3D shape meshes or other graph-structured data, to which traditional local convolution operators do not directly apply. To address this problem, we propose a novel graph-convolution operator to establish correspondences between filter weights and graph neighborhoods with arbitrary connectivity. The key novelty of our approach is that these correspondences are dynamically computed from features learned by the network, rather than relying on predefined static coordinates over the graph as in previous work. We obtain excellent experimental results that significantly improve over previous state-of-the-art shape correspondence results. This shows that our approach can learn effective shape representations from raw input coordinates, without relying on shape descriptors.
Document type :
Conference papers
Liste complète des métadonnées

Cited literature [35 references]  Display  Hide  Download


https://hal.inria.fr/hal-01540389
Contributor : Thoth Team <>
Submitted on : Wednesday, March 28, 2018 - 3:34:20 PM
Last modification on : Wednesday, March 20, 2019 - 1:29:03 AM
Document(s) archivé(s) le : Thursday, September 13, 2018 - 11:52:46 AM

Identifiers

Citation

Nitika Verma, Edmond Boyer, Jakob Verbeek. FeaStNet: Feature-Steered Graph Convolutions for 3D Shape Analysis. CVPR - IEEE Conference on Computer Vision & Pattern Recognition, Jun 2018, Salt Lake City, United States. pp.2598-2606, ⟨10.1109/CVPR.2018.00275⟩. ⟨hal-01540389v2⟩

Share

Metrics

Record views

994

Files downloads

1214