Abstract | ||
---|---|---|
This paper presents an end-to-end differentiable algorithm for anisotropic surface normal estimation on unstructured point-clouds. We utilize graph neural networks to iteratively infer point weights for a plane fitting algorithm applied to local neighborhoods. The approach retains the interpretability and efficiency of traditional sequential plane fitting while benefiting from a data-dependent deep-learning parameterization. This results in a state-of-the-art surface normal estimator that is robust to noise, outliers and point density variation and that preserves sharp features through anisotropic kernels and a local spatial transformer. Contrary to previous deep learning methods, the proposed approach does not require any hand-crafted features while being faster and more parameter efficient. |
Year | DOI | Venue |
---|---|---|
2019 | 10.1109/CVPR42600.2020.01126 | arXiv: Computer Vision and Pattern Recognition |
DocType | Volume | Citations |
Journal | abs/1904.07172 | 2 |
PageRank | References | Authors |
0.36 | 0 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Jan Eric Lenssen | 1 | 2 | 0.36 |
Christian Osendorfer | 2 | 125 | 13.24 |
Masci, Jonathan | 3 | 1158 | 82.31 |