DANC-Net: Dual-Attention and Negative Constraint Network for Point Cloud Classification

Convolutional neural networks, as a branch of deep neural networks, have been widely used in multidimensional signal processing, especially in point cloud signal processing. Nevertheless, in point cloud signal processing, most point cloud classification networks currently do not consider local featu...

Full description

Saved in:
Bibliographic Details
Main Authors: Hang Sun, Yuanyue Zhang, Jinmei Shi, Shuifa Sun, Guanqun Sheng, Yirong Wu
Format: Article
Language:English
Published: Wiley 2022-01-01
Series:International Journal of Antennas and Propagation
Online Access:http://dx.doi.org/10.1155/2022/5417440
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850230354729762816
author Hang Sun
Yuanyue Zhang
Jinmei Shi
Shuifa Sun
Guanqun Sheng
Yirong Wu
author_facet Hang Sun
Yuanyue Zhang
Jinmei Shi
Shuifa Sun
Guanqun Sheng
Yirong Wu
author_sort Hang Sun
collection DOAJ
description Convolutional neural networks, as a branch of deep neural networks, have been widely used in multidimensional signal processing, especially in point cloud signal processing. Nevertheless, in point cloud signal processing, most point cloud classification networks currently do not consider local feature correlation. In addition, they only adopt ground-truth as positive information to guide the training of networks while ignoring negative information. Therefore, this paper proposes a network model to classify point cloud signals based on feature correlation and negative constraint, DANC-Net (dual-attention and negative constraint on point cloud classification). In the DANC-Net, the dual-attention mechanism is utilized to strengthen the interaction between local features of point cloud signal from both channel and space, thereby improving the expression ability of extracted features. Moreover, during the training of the DANC-Net, the negative constraint loss function ensures that the features in the same categories are close and those in the different categories are far away from each other in the representation space, so as to improve the feature extraction capability of the network. Experiments demonstrate that the DANC-Net achieves better classification performance than the existing point cloud classification algorithms on synthetic datasets ModelNet10 and ModelNet40 and real-scene dataset ScanObjectNN. The code is released at https://github.com/sunhang1986/DANC-Net.
format Article
id doaj-art-2895d9e2d9e64a71b9fbe5922e9277ea
institution OA Journals
issn 1687-5877
language English
publishDate 2022-01-01
publisher Wiley
record_format Article
series International Journal of Antennas and Propagation
spelling doaj-art-2895d9e2d9e64a71b9fbe5922e9277ea2025-08-20T02:03:54ZengWileyInternational Journal of Antennas and Propagation1687-58772022-01-01202210.1155/2022/5417440DANC-Net: Dual-Attention and Negative Constraint Network for Point Cloud ClassificationHang Sun0Yuanyue Zhang1Jinmei Shi2Shuifa Sun3Guanqun Sheng4Yirong Wu5College of Computer and Information TechnologyCollege of Computer and Information TechnologyCollege of Information EngineeringCollege of Computer and Information TechnologyCollege of Computer and Information TechnologyCollege of Computer and Information TechnologyConvolutional neural networks, as a branch of deep neural networks, have been widely used in multidimensional signal processing, especially in point cloud signal processing. Nevertheless, in point cloud signal processing, most point cloud classification networks currently do not consider local feature correlation. In addition, they only adopt ground-truth as positive information to guide the training of networks while ignoring negative information. Therefore, this paper proposes a network model to classify point cloud signals based on feature correlation and negative constraint, DANC-Net (dual-attention and negative constraint on point cloud classification). In the DANC-Net, the dual-attention mechanism is utilized to strengthen the interaction between local features of point cloud signal from both channel and space, thereby improving the expression ability of extracted features. Moreover, during the training of the DANC-Net, the negative constraint loss function ensures that the features in the same categories are close and those in the different categories are far away from each other in the representation space, so as to improve the feature extraction capability of the network. Experiments demonstrate that the DANC-Net achieves better classification performance than the existing point cloud classification algorithms on synthetic datasets ModelNet10 and ModelNet40 and real-scene dataset ScanObjectNN. The code is released at https://github.com/sunhang1986/DANC-Net.http://dx.doi.org/10.1155/2022/5417440
spellingShingle Hang Sun
Yuanyue Zhang
Jinmei Shi
Shuifa Sun
Guanqun Sheng
Yirong Wu
DANC-Net: Dual-Attention and Negative Constraint Network for Point Cloud Classification
International Journal of Antennas and Propagation
title DANC-Net: Dual-Attention and Negative Constraint Network for Point Cloud Classification
title_full DANC-Net: Dual-Attention and Negative Constraint Network for Point Cloud Classification
title_fullStr DANC-Net: Dual-Attention and Negative Constraint Network for Point Cloud Classification
title_full_unstemmed DANC-Net: Dual-Attention and Negative Constraint Network for Point Cloud Classification
title_short DANC-Net: Dual-Attention and Negative Constraint Network for Point Cloud Classification
title_sort danc net dual attention and negative constraint network for point cloud classification
url http://dx.doi.org/10.1155/2022/5417440
work_keys_str_mv AT hangsun dancnetdualattentionandnegativeconstraintnetworkforpointcloudclassification
AT yuanyuezhang dancnetdualattentionandnegativeconstraintnetworkforpointcloudclassification
AT jinmeishi dancnetdualattentionandnegativeconstraintnetworkforpointcloudclassification
AT shuifasun dancnetdualattentionandnegativeconstraintnetworkforpointcloudclassification
AT guanqunsheng dancnetdualattentionandnegativeconstraintnetworkforpointcloudclassification
AT yirongwu dancnetdualattentionandnegativeconstraintnetworkforpointcloudclassification