Deep Parametric Continuous Convolutional Neural Networks

Abstract

Standard convolutional neural networks assume a grid structured input is available and exploit discrete convolutions as their fundamental building blocks. This limits their applicability to many real-world applications. In this paper we propose Parametric Continuous Convolution, a new learnable operator that operates over non-grid structured data. The key idea is to exploit parameterized kernel functions that span the full continuous vector space. This generalization allows us to learn over arbitrary data structures as long as their support relationship is computable. Our experiments show significant improvement over the state-of-the-art in point cloud segmentation of indoor and outdoor scenes, and lidar motion estimation of driving scenes.

Accepted by CVPR 2018

References in corpus (2)

• Geometric deep learning: going beyond Euclidean data
• SchNet: A continuous-filter convolutional neural network for modeling quantum interactions

Cited by in corpus (27)

• Deep Multi-modal Object Detection and Semantic Segmentation for Autonomous Driving: Datasets, Methods, and Challenges
• Deep Learning for Image and Point Cloud Fusion in Autonomous Driving: A Review
• Learning Semantic Segmentation of Large-Scale Point Clouds with Random Sampling
• PRA-Net: Point Relation-Aware Network for 3D Point Cloud Analysis
• JSENet: Joint Semantic Segmentation and Edge Detection Network for 3D Point Clouds
• A comprehensive overview of deep learning techniques for 3D point cloud classification and semantic segmentation
• Hierarchical Attention Learning of Scene Flow in 3D Point Clouds
• Deep Unsupervised Learning of 3D Point Clouds via Graph Topology Inference and Filtering
• Campus3D: A Photogrammetry Point Cloud Benchmark for Hierarchical Understanding of Outdoor Scene
• Advancements in Point Cloud-Based 3D Defect Detection and Classification for Industrial Systems: A Comprehensive Survey
• PVNAS: 3D Neural Architecture Search with Point-Voxel Convolution
• Dynamic Occupancy Grid Mapping with Recurrent Neural Networks
• Imbalance Knowledge-Driven Multi-modal Network for Land-Cover Semantic Segmentation Using Images and LiDAR Point Clouds
• Continuous Conditional Random Field Convolution for Point Cloud Segmentation
• Spatial Transformer for 3D Point Clouds
• Scan-based Semantic Segmentation of LiDAR Point Clouds: An Experimental Study
• Scale-covariant and scale-invariant Gaussian derivative networks
• GeoSegNet: Point Cloud Semantic Segmentation via Geometric Encoder-Decoder Modeling
• Unsupervised Learning of 3D Scene Flow from Monocular Camera
• Hyper-Convolution Networks for Biomedical Image Segmentation
• Push-the-Boundary: Boundary-aware Feature Propagation for Semantic Segmentation of 3D Point Clouds
• A Continuous Convolutional Trainable Filter for Modelling Unstructured Data
• Composite Convolution: a Flexible Operator for Deep Learning on 3D Point Clouds
• Neural Field Convolutions by Repeated Differentiation
• RMS-FlowNet++: Efficient and Robust Multi-Scale Scene Flow Estimation for Large-Scale Point Clouds
• DynaBench: A benchmark dataset for learning dynamical systems from low-resolution data
• Domain Adaptation-Based Crossmodal Knowledge Distillation for 3D Semantic Segmentation