Hierarchical Edge Aware Learning for 3D Point Cloud
Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review
Standard
Hierarchical Edge Aware Learning for 3D Point Cloud. / Li, Lei.
Advances in Computer Graphics - 40th Computer Graphics International Conference, CGI 2023, Proceedings. ed. / Bin Sheng; Lei Bi; Jinman Kim; Nadia Magnenat-Thalmann; Daniel Thalmann. Springer, 2024. p. 81-92 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 14495).Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review
Harvard
APA
Vancouver
Author
Bibtex
}
RIS
TY - GEN
T1 - Hierarchical Edge Aware Learning for 3D Point Cloud
AU - Li, Lei
N1 - Publisher Copyright: © 2024, The Author(s), under exclusive license to Springer Nature Switzerland AG.
PY - 2024
Y1 - 2024
N2 - This paper proposes an innovative approach to Hierarchical Edge Aware 3D Point Cloud Learning (HEA-Net) that seeks to address the challenges of noise in point cloud data, and improve object recognition and segmentation by focusing on edge features. In this study, we present an innovative edge-aware learning methodology, specifically designed to enhance point cloud classification and segmentation. Drawing inspiration from the human visual system, the concept of edge-awareness has been incorporated into this methodology, contributing to improved object recognition while simultaneously reducing computational time. Our research has led to the development of an advanced 3D point cloud learning framework that effectively manages object classification and segmentation tasks. A unique fusion of local and global network learning paradigms has been employed, enriched by edge-focused local and global embeddings, thereby significantly augmenting the model’s interpretative prowess. Further, we have applied a hierarchical transformer architecture to boost point cloud processing efficiency, thus providing nuanced insights into structural understanding. Our approach demonstrates significant promise in managing noisy point cloud data and highlights the potential of edge-aware strategies in 3D point cloud learning. The proposed approach is shown to outperform existing techniques in object classification and segmentation tasks, as demonstrated by experiments on ModelNet40 and ShapeNet datasets.
AB - This paper proposes an innovative approach to Hierarchical Edge Aware 3D Point Cloud Learning (HEA-Net) that seeks to address the challenges of noise in point cloud data, and improve object recognition and segmentation by focusing on edge features. In this study, we present an innovative edge-aware learning methodology, specifically designed to enhance point cloud classification and segmentation. Drawing inspiration from the human visual system, the concept of edge-awareness has been incorporated into this methodology, contributing to improved object recognition while simultaneously reducing computational time. Our research has led to the development of an advanced 3D point cloud learning framework that effectively manages object classification and segmentation tasks. A unique fusion of local and global network learning paradigms has been employed, enriched by edge-focused local and global embeddings, thereby significantly augmenting the model’s interpretative prowess. Further, we have applied a hierarchical transformer architecture to boost point cloud processing efficiency, thus providing nuanced insights into structural understanding. Our approach demonstrates significant promise in managing noisy point cloud data and highlights the potential of edge-aware strategies in 3D point cloud learning. The proposed approach is shown to outperform existing techniques in object classification and segmentation tasks, as demonstrated by experiments on ModelNet40 and ShapeNet datasets.
KW - 3D Point Cloud
KW - Classification
KW - Edge Learning
KW - Segmentation
UR - http://www.scopus.com/inward/record.url?scp=85184285841&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-50069-5_8
DO - 10.1007/978-3-031-50069-5_8
M3 - Article in proceedings
AN - SCOPUS:85184285841
SN - 9783031500688
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 81
EP - 92
BT - Advances in Computer Graphics - 40th Computer Graphics International Conference, CGI 2023, Proceedings
A2 - Sheng, Bin
A2 - Bi, Lei
A2 - Kim, Jinman
A2 - Magnenat-Thalmann, Nadia
A2 - Thalmann, Daniel
PB - Springer
T2 - 40th Computer Graphics International Conference, CGI 2023
Y2 - 28 August 2023 through 1 September 2023
ER -
ID: 385796082