SpiderCNN: Deep Learning on Point Sets with Parameterized Convolutional Filters
This addresses the problem of processing unstructured 3D data for computer vision applications, representing an incremental advance in point cloud deep learning.
The paper tackles the challenge of applying convolutional neural networks to irregular 3D point clouds by proposing SpiderCNN, a novel architecture that uses parameterized filters to extract geometric features, achieving state-of-the-art accuracy of 92.4% on ModelNet40 classification and competitive segmentation results.
Deep neural networks have enjoyed remarkable success for various vision tasks, however it remains challenging to apply CNNs to domains lacking a regular underlying structures such as 3D point clouds. Towards this we propose a novel convolutional architecture, termed SpiderCNN, to efficiently extract geometric features from point clouds. SpiderCNN is comprised of units called SpiderConv, which extend convolutional operations from regular grids to irregular point sets that can be embedded in R^n, by parametrizing a family of convolutional filters. We design the filter as a product of a simple step function that captures local geodesic information and a Taylor polynomial that ensures the expressiveness. SpiderCNN inherits the multi-scale hierarchical architecture from classical CNNs, which allows it to extract semantic deep features. Experiments on ModelNet40 demonstrate that SpiderCNN achieves state-of-the-art accuracy 92.4% on standard benchmarks, and shows competitive performance on segmentation task.