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

Abstract

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. https://github.com/lijx10/SO-Net

arXiv comment17 pages, CVPR 2018

References in corpus (8)

• Adam: A Method for Stochastic Optimization
• PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation
• Spectral Networks and Locally Connected Networks on Graphs
• PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space
• Generative and Discriminative Voxel Modeling with Convolutional Neural Networks
• Dynamic Edge-Conditioned Filters in Convolutional Neural Networks on Graphs
• Dominant Set Clustering and Pooling for Multi-View 3D Object Recognition
• Unsupervised Monocular Depth Estimation with Left-Right Consistency

Cited by in corpus (33)

• PointCNN: Convolution On $\mathcal{X}$-Transformed Points
• Directionally Constrained Fully Convolutional Neural Network For Airborne Lidar Point Cloud Classification
• Self-Contrastive Learning with Hard Negative Sampling for Self-supervised Point Cloud Learning
• Density-Aware Convolutional Networks with Context Encoding for Airborne LiDAR Point Cloud Classification
• STD: Sparse-to-Dense 3D Object Detector for Point Cloud
• Patch-based Progressive 3D Point Set Upsampling
• PCAN: 3D Attention Map Learning Using Contextual Information for Point Cloud Based Retrieval
• LPD-Net: 3D Point Cloud Learning for Large-Scale Place Recognition and Environment Analysis
• Microparticle cloud imaging and tracking for data-driven plasma science
• Coherent Point Drift Networks: Unsupervised Learning of Non-Rigid Point Set Registration
• Pointwise Rotation-Invariant Network with Adaptive Sampling and 3D Spherical Voxel Convolution
• Modeling Local Geometric Structure of 3D Point Clouds using Geo-CNN
• DAR-Net: Dynamic Aggregation Network for Semantic Scene Segmentation
• Deformable Filter Convolution for Point Cloud Reasoning
• JSNet: Joint Instance and Semantic Segmentation of 3D Point Clouds
• SAWNet: A Spatially Aware Deep Neural Network for 3D Point Cloud Processing
• Point2Node: Correlation Learning of Dynamic-Node for Point Cloud Feature Modeling
• Unsupervised 3D Learning for Shape Analysis via Multiresolution Instance Discrimination
• MeshNet: Mesh Neural Network for 3D Shape Representation
• Geometry Sharing Network for 3D Point Cloud Classification and Segmentation
• Point-to-Pose Voting based Hand Pose Estimation using Residual Permutation Equivariant Layer
• Improved Generalization Bounds of Group Invariant / Equivariant Deep Networks via Quotient Feature Spaces
• PVRNet: Point-View Relation Neural Network for 3D Shape Recognition
• Volumetric Convolution: Automatic Representation Learning in Unit Ball
• NPTC-net: Narrow-Band Parallel Transport Convolutional Neural Network on Point Clouds
• Point Transformer for Shape Classification and Retrieval of 3D and ALS Roof PointClouds
• LRC-Net: Learning Discriminative Features on Point Clouds by Encoding Local Region Contexts
• Local-Area-Learning Network: Meaningful Local Areas for Efficient Point Cloud Analysis
• Dense-Resolution Network for Point Cloud Classification and Segmentation
• VERAM: View-Enhanced Recurrent Attention Model for 3D Shape Classification
• Learning Equivariant Representations
• Zero-shot Learning of 3D Point Cloud Objects
• 3DMotion-Net: Learning Continuous Flow Function for 3D Motion Prediction