DiffusionNet: Discretization Agnostic Learning on Surfaces
About
We introduce a new general-purpose approach to deep learning on 3D surfaces, based on the insight that a simple diffusion layer is highly effective for spatial communication. The resulting networks are automatically robust to changes in resolution and sampling of a surface -- a basic property which is crucial for practical applications. Our networks can be discretized on various geometric representations such as triangle meshes or point clouds, and can even be trained on one representation then applied to another. We optimize the spatial support of diffusion as a continuous network parameter ranging from purely local to totally global, removing the burden of manually choosing neighborhood sizes. The only other ingredients in the method are a multi-layer perceptron applied independently at each point, and spatial gradient features to support directional filters. The resulting networks are simple, robust, and efficient. Here, we focus primarily on triangle mesh surfaces, and demonstrate state-of-the-art results for a variety of tasks including surface classification, segmentation, and non-rigid correspondence.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Mesh Saliency Prediction | Proposed Dataset Non-textured 1.0 (test) | CC0.4662 | 45 | |
| Mesh Saliency Prediction | Proposed Dataset Textured Mesh 1.0 (test) | CC0.4351 | 45 | |
| Non-isometric 3D shape matching | SMAL | Mean Geodesic Error6.6 | 22 | |
| Mesh Segmentation | Human Body dataset | Accuracy95.5 | 20 | |
| Near-isometric shape matching | SCAPE (final 20 shapes) | Pointwise Geodesic Error2.9 | 16 | |
| Near-isometric shape matching | FAUST (last 20 shapes) | Pointwise Geodesic Error2.6 | 16 | |
| Non-rigid shape matching | SCAPE | Mean Geodesic Error2.6 | 16 | |
| Saliency Prediction | SAL3D (test) | CC0.5371 | 15 | |
| Segmentation | Human Body 12k-vertex meshes | Accuracy90.3 | 14 | |
| Shape classification | SHREC-11 30-class | Accuracy99.4 | 14 |