SO-Net: Self-Organizing Network for Point Cloud Analysis

By Jiaxin Li, Ben M. Chen, Gim Hee Lee. 2018

1,011 citations recorded by the source corpus.

This paper presents SO-Net, a permutation invariant architecture for deep learning with orderless point clouds. The SO-Net models the spatial distribution of point cloud by building a Self-Organizing Map (SOM). Based on the SOM, SO-Net performs hierarchical feature extraction on individual points and SOM nodes, and ultimately represents the input point cloud by a single feature vector. The receptive field of the network can be systematically adjusted by conducting point-to-node k nearest neighbor search. In recognition tasks such as point cloud reconstruction, classification, object part segmentation and shape retrieval, our proposed network demonstrates performance that is similar with or better than state-of-the-art approaches. In addition, the training speed is significantly faster than existing point cloud recognition networks because of the parallelizability and simplicity of the proposed architecture. Our code is available at the project website.1

Subfields: Computational Mechanics, Environmental Engineering, 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 Environmental Engineering

In Geology

Citations in the corpus

Learning SO(3) Equivariant Representations with Spherical CNNs (2017, background): “…a neural network that operates directly on point clouds, which was followed by several others (Klokov and Lem- 1 The first version of this work was submitted to CVPR on 11/15/2017, shortly after we became aware of Cohen et al. (2018) ICLR submission on 10/27/2017. pitsky 2017; Li et al. 2018).”
PointCNN: Convolution On X-Transformed Points (2018, methodology): “…[55] 84.74 82.0 - - Pd-Network [22] 85.49 82.7 - - SSCN [12] 85.98 83.3 - - SPLATNet [43] 85.4 83.7 - - SpiderCNN [53] 85.3 81.7 - - SO-Net [27] 84.9 81.0 - - PCNN [3] 85.1 81.8 - - KCNet [42] 83.7 82.2 - - SpecGCN [46] 85.4 - - - Kd-Net [22] 82.3 77.4 - - 3DmFV-Net [4] 84.3 81.0 - - RSNet…”
Rotation Invariant Convolutions for 3D Point Clouds Deep Learning (2019, methodology): “Some methods exploit hierarchical structures and clustering for scalability [28, 15, 33, 34, 16], mapping point cloud to two dimensional space [40, 9, 18], applying spectral analysis [42], or addressing non-uniform point distribution [10].”
Point Transformer (2020, methodology): “Since current state-of-the-art methods have shown that aggregating local and global information increases the network’s capabilities of capturing context information [7], [17], [18], we employ SortNet to generate local features of the point cloud.”
Patch-Based Progressive 3D Point Set Upsampling (2018, background): “Following PointNet++ [42], most existing point-based networks extract multiple scales of information by hierarchically downsampling the input point sets [33, 59].”
Walk in the Cloud: Learning Curves for Point Clouds Shape Analysis (2021, methodology): “Following them, recent works have extended the point-wise method to various directions, which include designing advanced convolution operations [12, 43, 38, 20], considering a wider neighborhood [15, 50, 36, 22], and the adaptive aggregation of it [6, 44, 49, 45].”
PointASNL: Robust Point Clouds Processing Using Nonlocal Neural Networks With Adaptive Sampling (2020, methodology): “Then we compare our method with PointNet [24], PointNet++ [25], SO-Net [19] and the recent state-of-the-art RSCNN [22].”
Learning Object Bounding Boxes for 3D Instance Segmentation on Point Clouds (2019, background): “To address it, a number of approaches [38; 57; 43; 31; 55; 49; 26; 17] have been proposed recently.”
Self-Supervised Pretraining of 3D Features on any Point-Cloud (2021, methodology): “Most methods on self-supervised learning focus on single 3D object representation with different applications to reconstruction, classiﬁ-cation or part segmentation [1, 2, 25, 33, 35, 44, 51, 77, 103, 114].”
Mitigating the Hubness Problem for Zero-Shot Learning of 3D Objects (2019, methodology): “In the left branch, a feature vector κ(X ) ∈ Rm from a point cloud X is extracted using a point cloud network, which can be any network that learns a feature space from 3D point cloud data with the capability of being invariant to permutations in the point cloud [19, 26, 27, 37, 38, 43].”
3D Point Capsule Networks (2018, background): “making assumptions on the modality, point clouds are the preferred containers for many 3D applications [46,23]. Due to this widespread use, recent works such as PointNet [24], PointNet++ [25], SO-Net [20], spherical convolutions [18], Monte Carlo convolutions [10] and dynamic graph networks [42] hav”