Dynamic Graph CNN for Learning on Point Clouds
read the original abstract
Point clouds provide a flexible geometric representation suitable for countless applications in computer graphics; they also comprise the raw output of most 3D data acquisition devices. While hand-designed features on point clouds have long been proposed in graphics and vision, however, the recent overwhelming success of convolutional neural networks (CNNs) for image analysis suggests the value of adapting insight from CNN to the point cloud world. Point clouds inherently lack topological information so designing a model to recover topology can enrich the representation power of point clouds. To this end, we propose a new neural network module dubbed EdgeConv suitable for CNN-based high-level tasks on point clouds including classification and segmentation. EdgeConv acts on graphs dynamically computed in each layer of the network. It is differentiable and can be plugged into existing architectures. Compared to existing modules operating in extrinsic space or treating each point independently, EdgeConv has several appealing properties: It incorporates local neighborhood information; it can be stacked applied to learn global shape properties; and in multi-layer systems affinity in feature space captures semantic characteristics over potentially long distances in the original embedding. We show the performance of our model on standard benchmarks including ModelNet40, ShapeNetPart, and S3DIS.
This paper has not been read by Pith yet.
Forward citations
Cited by 9 Pith papers
-
Higgs Physics with the XFEL Compton $\boldsymbol{\gamma\gamma}$ Collider Concept at $\boldsymbol{\sqrt{s}=125}$ GeV
An XFEL Compton gamma-gamma collider at 125 GeV with a set transformer deep learning classifier on particle-flow point clouds can achieve high-precision Higgs measurements across hadronic, semi-leptonic, and leptonic ...
-
Neural Operator: Graph Kernel Network for Partial Differential Equations
Graph Kernel Networks learn PDE solution operators that generalize across discretization methods and grid resolutions using graph-based kernel integration.
-
Fast Graph Representation Learning with PyTorch Geometric
PyTorch Geometric is a PyTorch library that delivers fast graph neural network training through sparse GPU kernels and variable-size mini-batching.
-
Learning Embedding of 3D models with Quadric Loss
Quadric loss combined with Chamfer loss yields better sharp-feature reconstruction in 3D models than either loss alone or other point-surface alternatives.
-
From Full and Partial Intraoral Scans to Crown Proposal: A Classification-Guided Restoration Assistance Pipeline
A classification-routed pipeline segments partial and full intraoral scans then retrieves and fits crown proposals from neighboring teeth embeddings, reporting macro DSC 0.9249 on 1958 partial scans.
-
Explainable AI for Jet Tagging: A Comparative Study of GNNExplainer, GNNShap, and GradCAM for Jet Tagging in the Lund Jet Plane
Explainability techniques applied to LundNet show that assigned node importance correlates with classical jet substructure observables such as N-subjettiness ratios and energy correlation functions, with shifts across...
-
E-PCN: Jet Tagging with Explainable Particle Chebyshev Networks Using Kinematic Features
E-PCN reaches 94.67% macro-accuracy on 10-class jet tagging by weighting graphs with angular separation, transverse momentum, momentum fraction, and invariant mass, with Grad-CAM showing the first two account for 76% ...
-
Mitigating the Hubness Problem for Zero-Shot Learning of 3D Objects
A specialized loss mitigates hubness bias in 3D zero-shot learning and sets new state-of-the-art results on ModelNet40, ModelNet10, McGill, and SHREC2015 for both ZSL and GZSL.
-
How much real data do we actually need: Analyzing object detection performance using synthetic and real data
Synthetic data can partially substitute for real data in object detection training, with performance tied to domain similarity and the volume of real data included.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.