SpiderCNN: Deep Learning on Point Sets with Parameterized Convolutional Filters

By Yifan Xu, Tianqi Fan, Mingye Xu, Long Zeng, Yu Qiao. 2018

867 citations recorded by the source corpus.

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 set that can be embedded in R^n, by parametrizing a family of convolutional filters. We elaborately design the filter as a product of simple step function that captures local geodesic information and a Taylor polynomial that ensures the expressiveness. SpiderCNN inherits the multi-scale hierarchical architecture from the classical CNNs, which allows it to extract semantic deep features. Experiments on ModelNet40 demonstrate that SpiderCNN achieves the-state-of-the art accuracy 92.4% on standard benchmarks, and shows competitive performance on segmentation task.

Subfields: Computational Mechanics, Computer Vision and Pattern Recognition, Geology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Mechanics

In Computer Vision and Pattern Recognition

In Geology

Citations in the corpus

SE(3)-Transformers: 3D Roto-Translation Equivariant Attention Networks (2020, background): “DeepSet[30] 3DmFV[3] Set Transformer[12] PointNet[15] SpiderCNN[29] Tensor Field +z[20] PointNet++[16] SE(3)-Transf.”
Revisiting Point Cloud Classification: A New Benchmark Dataset and Classification Model on Real-World Data (2019, methodology): “It can be explained by the fact that (c) SpiderCNN [44], (d) PointNet++ [32], (e) DGCNN [42] and (f) PointCNN [25] on our hardest PB T50 RS.”
PointNeXt: Revisiting PointNet++ with Improved Training and Scaling Strategies (2022, neutral): “[57, 21, 47, 46] project point clouds onto pseudo grids to allow for regular convolutions.”
View-GCN: View-Based Graph Convolutional Network for 3D Shape Analysis (2020, positive): “Succeeding methods, e.g ., PointNet++ [42], Spi-derCNN [52], PointCNN [32], RS-CNN [33] achieve improved performance for 3D shape recognition.”
Point Transformer (2020, background): “[30] Y. Xu, T. Fan, M. Xu, L. Zeng, and Y. Qiao, ‘‘SpiderCNN: Deep learning on point sets with parameterized convolutional filters,’’ in Proc.”
PointContrast: Unsupervised Pre-training for 3D Point Cloud Understanding (2020, background): “Second, is the choice of neighborhood aggregation mechanism which could either be hierarchical [48, 33, 82, 16, 35], spatial CNN-like [30, 73, 37, 85, 59], spectral [78, 62, 65] or graph-based [72, 64, 68, 54].”
Global-Local Bidirectional Reasoning for Unsupervised Representation Learning of 3D Point Clouds (2020, background): “To tackle this problem, current state-of-the-arts on point cloud analysis [2,26,28,33,38,43,49,51,54] are established with the help of extensive human-annotated supervised information.”
Rethinking Network Design and Local Geometry in Point Cloud: A Simple Residual MLP Framework (2022, neutral): “(Xu et al., 2018) 69.8 73.7 PointNet++ (Qi et al., 2017b) 75.4 77.9 DGCNN (Wang et al., 2019) 73.6 78.1 PointCNN (Li et al., 2018b) 75.1 78.5 BGA-DGCNN (Uy et al., 2019) 75.7 79.7 BGA-PN++ (Uy et al., 2019) 77.5 80.2 DRNet (Qiu et al., 2021a) 78.0 80.3 GBNet (Qiu et al., 2021b) 77.8 80.5 SimpleView…”
Language-Grounded Indoor 3D Semantic Segmentation in the Wild (2022, methodology): “The introduction of PointNet [41] presented a point-based alternative with strong memory efficiency by operating on unstructured point clouds, with various methods introducing local operators to better learn neighborhood structures [42,48,53].”
Learning Object Bounding Boxes for 3D Instance Segmentation on Point Clouds (2019, methodology): “To address it, a number of approaches [38; 57; 43; 31; 55; 49; 26; 17] have been proposed recently.”
Hopfield Networks is All You Need (2020, background): “Possible applications for a Hopfield layer in a deep network architecture comprise: • multiple instance learning (MIL) [26], • processing of and learning with point sets [51, 52, 73], • set-based and permutation invariant learning [32, 55, 76, 40, 38, 78], • attention-based learning [64], • associat”
Point2Sequence: Learning the Shape Representation of 3D Point Clouds with an Attention-based Sequence to Sequence Network (2018, background): “As a pioneer, PointNet (Qi et al. 2017b) achieves satisfactory performance (Xu et al. 2018) investigate the convolution-like operations which aggregate the neighbors of a given point by edge attributes in the local region graph.”
Masked Autoencoders for Point Cloud Self-supervised Learning (2022, neutral): “OBJ-BG OBJ-ONLY PB-T50-RS PointNet [29] 73.3 79.2 68.0 SpiderCNN [50] 77.1 79.5 73.7 PointNet++ [30] 82.3 84.3 77.9 DGCNN [44] 82 We conduct experiments on three variants: OBJ-BG, OBJ-ONLY, and PB-T50-RS.”
Lagrangian Fluid Simulation with Continuous Convolutions (2020, methodology): “A number of methods transferred the convolution concept to point clouds in the context of semantic classification and segmentation of 3D objects (Hua et al., 2018; Atzmon et al., 2018; Hermosilla et al., 2018; Li et al., 2018; Su et al., 2018; Wu et al., 2019; Xu et al., 2018; Lei et al., 2019).”