MDSSN: An end-to-end deep network on triangle mesh parameterization

Article Computer Science, Software Engineering

Mesh Convolutional Networks With Face and Vertex Feature Operators

Daniel Perez et al.

Summary: In this study, we propose face-based and vertex-based operators for mesh convolutional networks. We design two novel architectures based on the MeshCNN network that can operate on faces and vertices of a mesh, respectively. The proposed face-based architecture outperforms the original MeshCNN implementation in mesh classification and mesh segmentation, setting the new state of the art on benchmark datasets. Additionally, the vertex-based operator is extended to the Point2Mesh model for mesh reconstruction, reducing training and inference time significantly.

IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS (2023)

Add to Collection

Article Computer Science, Information Systems

Voxel Structure-Based Mesh Reconstruction From a 3D Point Cloud

Chenlei Lv et al.

Summary: This paper proposes a voxel structure-based mesh reconstruction framework to improve the accuracy of local region detection. The experiment results demonstrate that our framework has significant advantages over peer methods in terms of mesh quality, geometric feature keeping, and processing speed.

IEEE TRANSACTIONS ON MULTIMEDIA (2022)

Add to Collection

Article Computer Science, Artificial Intelligence

DMMAN: A two-stage audio-visual fusion framework for sound separation and event localization

Ruihan Hu et al.

Summary: This paper introduces the Deep Multi-Modal Attention Network (DMMAN) model, which tackles sound separation and modal synchronization issues through a multi-modal separator and multi-modal matching classifier module. The model uses a two-stage fusion of sound and visual features, employing regression and classification losses to construct the DMMAN's loss function, leading to sound source and event localization tasks.

NEURAL NETWORKS (2021)

Add to Collection

Article Computer Science, Software Engineering

DNF-Net: A Deep Normal Filtering Network for Mesh Denoising

Xianzhi Li et al.

Summary: DNF-Net is a deep normal filtering network for mesh denoising that utilizes local patches to capture geometry information, does not require manual feature extraction, and shows superior denoising performance compared to existing techniques.

IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS (2021)

Add to Collection

Article Computer Science, Software Engineering

HodgeNet: Learning Spectral Geometry on Triangle Meshes

Dmitriy Smirnov et al.

Summary: This study introduces a technique for learning from meshes built from standard geometry processing modules and operations, which efficiently calculates eigenvalues/eigenvectors and generates spectral features similar to classical descriptors.

ACM TRANSACTIONS ON GRAPHICS (2021)

Add to Collection

Article Computer Science, Information Systems

Mesh Convolution: A Novel Feature Extraction Method for 3D Nonrigid Object Classification

Yu Chen et al.

Summary: In this paper, a novel feature extraction method called 3D-MConv is proposed to improve 3D nonrigid shape analysis by extending convolution operations to irregular mesh sets. The method can better handle global isometric deformation and articulation. Experimental results demonstrate that the proposed method achieves state-of-the-art accuracy on standard benchmarks.

IEEE TRANSACTIONS ON MULTIMEDIA (2021)

Add to Collection

Article Computer Science, Artificial Intelligence

3D Mesh Model Classification with a Capsule Network

Yang Zheng et al.

Summary: This paper proposes a deep learning method based on capsule network for three-dimensional mesh data classification. By designing a polynomial convolution template and improving the capsule network attitude parameter learning method, the classification effect of 3D mesh models is effectively enhanced.

ALGORITHMS (2021)

Add to Collection

Article Computer Science, Software Engineering