Point-GSMAE: A graph convolution and scale-based masked autoencoder for 3D point cloud representation

Masked Autoencoders (MAEs) have demonstrated considerable potential in advancing self-supervised learning for 3D point cloud representation. Nevertheless, existing MAE-based approaches, predominantly relying on Transformer architectures, struggle to effectively model interactions between points in l...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Information sciences Jg. 719; S. 122474
Hauptverfasser: Bai, Yun, Yang, Chaozhi, Li, Guanlin, He, Xiao, Xiao, Qian, Li, Zongmin
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Elsevier Inc 01.11.2025
Schlagworte:
ISSN:0020-0255
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Abstract Masked Autoencoders (MAEs) have demonstrated considerable potential in advancing self-supervised learning for 3D point cloud representation. Nevertheless, existing MAE-based approaches, predominantly relying on Transformer architectures, struggle to effectively model interactions between points in local neighborhoods. This limitation hinders the ability to capture fine-grained local geometric structures of point clouds, negatively impacting tasks that depend on local geometric relationships. In response to this issue, we present Point-GSMAE, an innovative MAE framework for 3D point clouds. This framework integrates graph convolution and graph scale to enhance local geometric modeling. Specifically, it constructs weighted adjacency matrices to encode relationships between neighboring points, edges, and center points, enabling graph convolution to aggregate neighborhood information into center points for precise modeling of local geometric structures. Furthermore, we introduce a graph scale component as a complementary descriptor capturing both the graph structure and spatial distribution, enriching the representation of local geometric properties. To further refine the learned representations, we incorporate a scale consistency loss function, aligning the reconstructed point clouds with their original structures and improving sensitivity to scale variations within local neighborhoods. Comprehensive experiments on multiple datasets demonstrate the efficacy of Point-GSMAE, outperforming existing Transformer-based MAE methods while requiring fewer parameters.
AbstractList Masked Autoencoders (MAEs) have demonstrated considerable potential in advancing self-supervised learning for 3D point cloud representation. Nevertheless, existing MAE-based approaches, predominantly relying on Transformer architectures, struggle to effectively model interactions between points in local neighborhoods. This limitation hinders the ability to capture fine-grained local geometric structures of point clouds, negatively impacting tasks that depend on local geometric relationships. In response to this issue, we present Point-GSMAE, an innovative MAE framework for 3D point clouds. This framework integrates graph convolution and graph scale to enhance local geometric modeling. Specifically, it constructs weighted adjacency matrices to encode relationships between neighboring points, edges, and center points, enabling graph convolution to aggregate neighborhood information into center points for precise modeling of local geometric structures. Furthermore, we introduce a graph scale component as a complementary descriptor capturing both the graph structure and spatial distribution, enriching the representation of local geometric properties. To further refine the learned representations, we incorporate a scale consistency loss function, aligning the reconstructed point clouds with their original structures and improving sensitivity to scale variations within local neighborhoods. Comprehensive experiments on multiple datasets demonstrate the efficacy of Point-GSMAE, outperforming existing Transformer-based MAE methods while requiring fewer parameters.
ArticleNumber 122474
Author Yang, Chaozhi
He, Xiao
Li, Guanlin
Xiao, Qian
Li, Zongmin
Bai, Yun
Author_xml – sequence: 1
  givenname: Yun
  surname: Bai
  fullname: Bai, Yun
  organization: School of Mathematics and Computer Science, Lishui University, Lishui, 323000, China
– sequence: 2
  givenname: Chaozhi
  surname: Yang
  fullname: Yang, Chaozhi
  organization: QINGDAO Institute of Software, College of Computer Science and Technology, China University of Petroleum (East China), Shandong, 266580, China
– sequence: 3
  givenname: Guanlin
  orcidid: 0000-0003-0774-2878
  surname: Li
  fullname: Li, Guanlin
  organization: QINGDAO Institute of Software, College of Computer Science and Technology, China University of Petroleum (East China), Shandong, 266580, China
– sequence: 4
  givenname: Xiao
  surname: He
  fullname: He, Xiao
  organization: QINGDAO Institute of Software, College of Computer Science and Technology, China University of Petroleum (East China), Shandong, 266580, China
– sequence: 5
  givenname: Qian
  surname: Xiao
  fullname: Xiao, Qian
  organization: QINGDAO Institute of Software, College of Computer Science and Technology, China University of Petroleum (East China), Shandong, 266580, China
– sequence: 6
  givenname: Zongmin
  surname: Li
  fullname: Li, Zongmin
  email: lizongmin@upc.edu.cn
  organization: QINGDAO Institute of Software, College of Computer Science and Technology, China University of Petroleum (East China), Shandong, 266580, China
BookMark eNp9kMtOwzAQRb0oEi3wAez8Awl-xUlgVZVSkIpAAtaWY0_AJbUjO63E35OorFldaaR7dOcs0MwHDwhdU5JTQuXNLnc-5YywIqeMiVLM0JwQRrLxUpyjRUo7QogopZwjeA3OD9nm7Xm5vsVL_Bl1_4VN8MfQHQYXPNbe4mR0B1mjE1i81-l7DH0YAngTLETchoj5Pe4nFDZdOFgcoY-QwA96glyis1Z3Ca7-8gJ9PKzfV4_Z9mXztFpuM8MKOmSCGcq1FNrqsmk5LYyF2lTEVnVb1awxohatoVRQLiUvrLbclppaIKKSWjb8AtET18SQUoRW9dHtdfxRlKjJjdqp0Y2a3KiTm7Fzd-rAOOzoIKpk3PgZWBfBDMoG90_7F_sHcSs
Cites_doi 10.1016/j.ins.2022.10.135
10.1016/j.inffus.2023.102043
10.1109/TMM.2021.3074240
10.1109/TIP.2014.2311377
10.1016/j.cag.2019.11.005
10.1109/JAS.2023.123432
ContentType Journal Article
Copyright 2025 Elsevier Inc.
Copyright_xml – notice: 2025 Elsevier Inc.
DBID AAYXX
CITATION
DOI 10.1016/j.ins.2025.122474
DatabaseName CrossRef
DatabaseTitle CrossRef
DatabaseTitleList
DeliveryMethod fulltext_linktorsrc
Discipline Engineering
Library & Information Science
ExternalDocumentID 10_1016_j_ins_2025_122474
S0020025525006061
GroupedDBID --K
--M
--Z
-~X
.DC
.~1
0R~
1B1
1OL
1RT
1~.
1~5
29I
4.4
457
4G.
5GY
5VS
7-5
71M
77I
8P~
9JN
9JO
AAAKF
AAAKG
AABNK
AAEDT
AAEDW
AAIKJ
AAKOC
AALRI
AAOAW
AAQFI
AAQXK
AARIN
AATTM
AAXKI
AAXUO
AAYFN
AAYWO
ABAOU
ABBOA
ABEFU
ABFNM
ABJNI
ABMAC
ABUCO
ABWVN
ABXDB
ACDAQ
ACGFS
ACLOT
ACNNM
ACRLP
ACRPL
ACVFH
ACZNC
ADBBV
ADCNI
ADEZE
ADGUI
ADJOM
ADMUD
ADNMO
ADTZH
ADVLN
AEBSH
AECPX
AEIPS
AEKER
AENEX
AEUPX
AFFNX
AFJKZ
AFPUW
AFTJW
AGHFR
AGQPQ
AGUBO
AGYEJ
AHHHB
AHJVU
AHZHX
AIALX
AIEXJ
AIGII
AIGVJ
AIIUN
AIKHN
AITUG
AKBMS
AKRWK
AKYEP
ALMA_UNASSIGNED_HOLDINGS
AMRAJ
ANKPU
AOUOD
APLSM
APXCP
ARUGR
ASPBG
AVWKF
AXJTR
AZFZN
BJAXD
BKOJK
BLXMC
CS3
DU5
EBS
EFJIC
EFKBS
EFLBG
EJD
EO8
EO9
EP2
EP3
F5P
FDB
FEDTE
FGOYB
FIRID
FNPLU
FYGXN
G-Q
GBLVA
GBOLZ
HAMUX
HLZ
HVGLF
HZ~
H~9
IHE
J1W
JJJVA
KOM
LG9
LY1
M41
MHUIS
MO0
MS~
N9A
O-L
O9-
OAUVE
OZT
P-8
P-9
P2P
PC.
Q38
R2-
ROL
RPZ
SBC
SDF
SDG
SDP
SDS
SES
SEW
SPC
SPCBC
SSB
SSD
SST
SSV
SSW
SSZ
T5K
TN5
TWZ
UHS
WH7
WUQ
XPP
YYP
ZMT
ZY4
~02
~G-
~HD
9DU
AAYXX
CITATION
ID FETCH-LOGICAL-c251t-42c13a64ada7bf315cde9c80d89f892bc494fc114136635dad3d7a1de0486a6b3
ISICitedReferencesCount 0
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001528665100001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 0020-0255
IngestDate Sat Nov 29 07:35:33 EST 2025
Sat Oct 11 16:51:56 EDT 2025
IsPeerReviewed true
IsScholarly true
Keywords Graph scale
Point cloud
Graph convolution
Masked autoencoder
Language English
LinkModel OpenURL
MergedId FETCHMERGED-LOGICAL-c251t-42c13a64ada7bf315cde9c80d89f892bc494fc114136635dad3d7a1de0486a6b3
ORCID 0000-0003-0774-2878
ParticipantIDs crossref_primary_10_1016_j_ins_2025_122474
elsevier_sciencedirect_doi_10_1016_j_ins_2025_122474
PublicationCentury 2000
PublicationDate November 2025
2025-11-00
PublicationDateYYYYMMDD 2025-11-01
PublicationDate_xml – month: 11
  year: 2025
  text: November 2025
PublicationDecade 2020
PublicationTitle Information sciences
PublicationYear 2025
Publisher Elsevier Inc
Publisher_xml – name: Elsevier Inc
References Wang, Huang, Hou, Zhang, Shan (br0310) 2019
Qi, Dong, Fan, Ge, Zhang, Ma, Yi (br0360) 2023
Yu, Rui, Tao (br0190) 2014; 23
Qi, Su, Mo, Guibas (br0010) 2017
Zhao, Jiang, Jia, Torr, Koltun (br0080) 2021
Qi, Yi, Su, Guibas (br0020) 2017; 30
Chang, Funkhouser, Guibas, Hanrahan, Huang, Li, Savarese, Savva, Song, Su (br0380) 2015
Wang, Sun, Liu, Sarma, Bronstein, Solomon (br0030) 2019; 38
Thomas, Qi, Deschaud, Marcotegui, Goulette, Guibas (br0240) 2019
Wu, Song, Khosla, Yu, Zhang, Tang, Xiao (br0400) 2015
Dong, Qi, Zhang, Zhang, Sun, Ge, Yi, Ma (br0210) 2022
Uy, Pham, Hua, Nguyen, Yeung (br0390) 2019
Qiu, Anwar, Barnes (br0070) 2021; 24
Wen, Long, Yu, Tao (br0060) 2024
Guo, Zhang, Qiu, Li, Heng (br0370) 2023
Zhou, Feng, Fang, Wei, Qin, Lu (br0040) 2021
Liu, Cai, Lee (br0430) 2022
Ma, Qin, You, Ran, Fu (br0420) 2022
Hamdi, Giancola, Ghanem (br0270) 2021
Afham, Dissanayake, Dissanayake, Dharmasiri, Thilakarathna, Rodrigo (br0330) 2022
Liu, Tian, Lv, Li, Wang (br0090) 2023; 11
Xu, Ding, Zhao, Qi (br0250) 2021
Goyal, Law, Liu, Newell, Deng (br0410) 2021
Zha, Ji, Li, Li, Dai, Chen, Wang, Xia (br0100) 2024; vol. 38
Graham, Engelcke, Van Der Maaten (br0260) 2018
Chen, Wang, Yang, Yu, Yuan, Yue (br0440) 2024; 36
Tian, Zheng, Zuo, Zhang, Zhang, Lin (br0170) 2024; 102
Li, Muller, Thabet, Ghanem (br0230) 2019
Armeni, Sener, Zamir, Jiang, Brilakis, Fischer, Savarese (br0450) 2016
Sanghi (br0340) 2020
Xie, Wang, Yu, Anandkumar, Alvarez, Luo (br0150) 2021; 34
Howard, Sandler, Chu, Chen, Chen, Tan, Wang, Zhu, Pang, Vasudevan (br0130) 2019
Wang, Zha, He, Yang, Zhang (br0350) 2024
Redmon, Divvala, Girshick, Farhadi (br0140) 2016
Yu, Tang, Rao, Huang, Zhou, Lu (br0280) 2022
Zhang, Guo, Gao, Fang, Zhao, Wang, Qiao, Li (br0200) 2022; 35
Zhang, Wang, Qiao, Gao, Li (br0220) 2023
Pang, Wang, Tay, Liu, Tian, Yuan (br0120) 2022
Van Nguyen, Lai, Veyseh, Nguyen (br0160) 2021
Te, Hu, Zheng, Guo (br0050) 2018
Zhang, Rabbat (br0290) 2018
Xie, Gu, Guo, Qi, Guibas, Litany (br0320) 2020
Zhang, Zhang, Yan (br0110) 2024
Chen, Song, Su, Zhao, Zhang (br0180) 2022; 616
Lu, Chen, Xie, Luo (br0300) 2020; 86
Yu (10.1016/j.ins.2025.122474_br0280) 2022
Liu (10.1016/j.ins.2025.122474_br0430) 2022
Wang (10.1016/j.ins.2025.122474_br0350) 2024
Graham (10.1016/j.ins.2025.122474_br0260) 2018
Hamdi (10.1016/j.ins.2025.122474_br0270) 2021
Wu (10.1016/j.ins.2025.122474_br0400) 2015
Thomas (10.1016/j.ins.2025.122474_br0240) 2019
Uy (10.1016/j.ins.2025.122474_br0390) 2019
Guo (10.1016/j.ins.2025.122474_br0370)
Li (10.1016/j.ins.2025.122474_br0230) 2019
Ma (10.1016/j.ins.2025.122474_br0420)
Dong (10.1016/j.ins.2025.122474_br0210)
Wang (10.1016/j.ins.2025.122474_br0030) 2019; 38
Liu (10.1016/j.ins.2025.122474_br0090) 2023; 11
Xu (10.1016/j.ins.2025.122474_br0250) 2021
Afham (10.1016/j.ins.2025.122474_br0330) 2022
Goyal (10.1016/j.ins.2025.122474_br0410) 2021
Van Nguyen (10.1016/j.ins.2025.122474_br0160)
Zhang (10.1016/j.ins.2025.122474_br0220) 2023
Xie (10.1016/j.ins.2025.122474_br0150) 2021; 34
Te (10.1016/j.ins.2025.122474_br0050) 2018
Sanghi (10.1016/j.ins.2025.122474_br0340) 2020
Qiu (10.1016/j.ins.2025.122474_br0070) 2021; 24
Lu (10.1016/j.ins.2025.122474_br0300) 2020; 86
Zhou (10.1016/j.ins.2025.122474_br0040) 2021
Qi (10.1016/j.ins.2025.122474_br0010) 2017
Zhang (10.1016/j.ins.2025.122474_br0200) 2022; 35
Pang (10.1016/j.ins.2025.122474_br0120) 2022
Chen (10.1016/j.ins.2025.122474_br0180) 2022; 616
Wang (10.1016/j.ins.2025.122474_br0310) 2019
Redmon (10.1016/j.ins.2025.122474_br0140) 2016
Xie (10.1016/j.ins.2025.122474_br0320) 2020
Chang (10.1016/j.ins.2025.122474_br0380)
Chen (10.1016/j.ins.2025.122474_br0440) 2024; 36
Armeni (10.1016/j.ins.2025.122474_br0450) 2016
Qi (10.1016/j.ins.2025.122474_br0020) 2017; 30
Zha (10.1016/j.ins.2025.122474_br0100) 2024; vol. 38
Zhang (10.1016/j.ins.2025.122474_br0290) 2018
Howard (10.1016/j.ins.2025.122474_br0130) 2019
Tian (10.1016/j.ins.2025.122474_br0170) 2024; 102
Zhao (10.1016/j.ins.2025.122474_br0080) 2021
Qi (10.1016/j.ins.2025.122474_br0360) 2023
Wen (10.1016/j.ins.2025.122474_br0060) 2024
Yu (10.1016/j.ins.2025.122474_br0190) 2014; 23
Zhang (10.1016/j.ins.2025.122474_br0110)
References_xml – start-page: 1314
  year: 2019
  end-page: 1324
  ident: br0130
  article-title: Searching for mobilenetv3
  publication-title: Proceedings of the IEEE/CVF International Conference on Computer Vision
– volume: 36
  year: 2024
  ident: br0440
  article-title: Pointgpt: auto-regressively generative pre-training from point clouds
  publication-title: Adv. Neural Inf. Process. Syst.
– start-page: 1912
  year: 2015
  end-page: 1920
  ident: br0400
  article-title: 3d shapenets: a deep representation for volumetric shapes
  publication-title: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition
– start-page: 10296
  year: 2019
  end-page: 10305
  ident: br0310
  article-title: Graph attention convolution for point cloud semantic segmentation
  publication-title: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition
– start-page: 19313
  year: 2022
  end-page: 19322
  ident: br0280
  article-title: Point-bert: pre-training 3d point cloud transformers with masked point modeling
  publication-title: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition
– year: 2022
  ident: br0210
  article-title: Autoencoders as cross-modal teachers: can pretrained 2d image transformers help 3d representation learning?
– start-page: 1588
  year: 2019
  end-page: 1597
  ident: br0390
  article-title: Revisiting point cloud classification: a new benchmark dataset and classification model on real-world data
  publication-title: Proceedings of the IEEE/CVF International Conference on Computer Vision
– start-page: 746
  year: 2018
  end-page: 754
  ident: br0050
  article-title: Rgcnn: regularized graph cnn for point cloud segmentation
  publication-title: Proceedings of the 26th ACM International Conference on Multimedia
– year: 2022
  ident: br0420
  article-title: Rethinking network design and local geometry in point cloud: a simple residual mlp framework
– year: 2024
  ident: br0350
  article-title: Rethinking masked representation learning for 3d point cloud understanding
  publication-title: IEEE Trans. Image Process.
– year: 2023
  ident: br0370
  article-title: Joint-mae: 2d-3d joint masked autoencoders for 3d point cloud pre-training
– start-page: 657
  year: 2022
  end-page: 675
  ident: br0430
  article-title: Masked discrimination for self-supervised learning on point clouds
  publication-title: European Conference on Computer Vision
– volume: 35
  start-page: 27061
  year: 2022
  end-page: 27074
  ident: br0200
  article-title: Point-m2ae: multi-scale masked autoencoders for hierarchical point cloud pre-training
  publication-title: Adv. Neural Inf. Process. Syst.
– start-page: 1534
  year: 2016
  end-page: 1543
  ident: br0450
  article-title: 3d semantic parsing of large-scale indoor spaces
  publication-title: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition
– start-page: 604
  year: 2022
  end-page: 621
  ident: br0120
  article-title: Masked autoencoders for point cloud self-supervised learning
  publication-title: European Conference on Computer Vision
– start-page: 779
  year: 2016
  end-page: 788
  ident: br0140
  article-title: You only look once: unified, real-time object detection
  publication-title: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition
– start-page: 21769
  year: 2023
  end-page: 21780
  ident: br0220
  article-title: Learning 3d representations from 2d pre-trained models via image-to-point masked autoencoders
  publication-title: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition
– volume: 102
  year: 2024
  ident: br0170
  article-title: A cross transformer for image denoising
  publication-title: Inf. Fusion
– start-page: 652
  year: 2017
  end-page: 660
  ident: br0010
  article-title: Pointnet: deep learning on point sets for 3d classification and segmentation
  publication-title: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition
– start-page: 574
  year: 2020
  end-page: 591
  ident: br0320
  article-title: Pointcontrast: unsupervised pre-training for 3d point cloud understanding
  publication-title: Computer Vision–ECCV 2020: 16th European Conference
– start-page: 3173
  year: 2021
  end-page: 3182
  ident: br0250
  article-title: Paconv: position adaptive convolution with dynamic kernel assembling on point clouds
  publication-title: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition
– volume: 616
  start-page: 526
  year: 2022
  end-page: 538
  ident: br0180
  article-title: Learning user sentiment orientation in social networks for sentiment analysis
  publication-title: Inf. Sci.
– start-page: 6411
  year: 2019
  end-page: 6420
  ident: br0240
  article-title: Kpconv: flexible and deformable convolution for point clouds
  publication-title: Proceedings of the IEEE/CVF International Conference on Computer Vision
– start-page: 4965
  year: 2021
  end-page: 4974
  ident: br0040
  article-title: Adaptive graph convolution for point cloud analysis
  publication-title: Proceedings of the IEEE/CVF International Conference on Computer Vision
– volume: 86
  start-page: 42
  year: 2020
  end-page: 51
  ident: br0300
  article-title: Pointngcnn: deep convolutional networks on 3d point clouds with neighborhood graph filters
  publication-title: Comput. Graph.
– volume: 30
  year: 2017
  ident: br0020
  article-title: Pointnet++: deep hierarchical feature learning on point sets in a metric space
  publication-title: Adv. Neural Inf. Process. Syst.
– volume: 34
  start-page: 12077
  year: 2021
  end-page: 12090
  ident: br0150
  article-title: Segformer: simple and efficient design for semantic segmentation with transformers
  publication-title: Adv. Neural Inf. Process. Syst.
– year: 2024
  ident: br0110
  article-title: Pcp-mae: learning to predict centers for point masked autoencoders
– start-page: 9267
  year: 2019
  end-page: 9276
  ident: br0230
  article-title: Deepgcns: can gcns go as deep as cnns?
  publication-title: Proceedings of the IEEE/CVF International Conference on Computer Vision
– year: 2021
  ident: br0160
  article-title: Trankit: a light-weight transformer-based toolkit for multilingual natural language processing
– start-page: 3809
  year: 2021
  end-page: 3820
  ident: br0410
  article-title: Revisiting point cloud shape classification with a simple and effective baseline
  publication-title: International Conference on Machine Learning
– volume: 11
  start-page: 231
  year: 2023
  end-page: 239
  ident: br0090
  article-title: Point cloud classification using content-based transformer via clustering in feature space
  publication-title: IEEE/CAA J. Autom. Sin.
– volume: 23
  start-page: 2019
  year: 2014
  end-page: 2032
  ident: br0190
  article-title: Click prediction for web image reranking using multimodal sparse coding
  publication-title: IEEE Trans. Image Process.
– start-page: 9902
  year: 2022
  end-page: 9912
  ident: br0330
  article-title: Crosspoint: self-supervised cross-modal contrastive learning for 3d point cloud understanding
  publication-title: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition
– start-page: 6279
  year: 2018
  end-page: 6283
  ident: br0290
  article-title: A graph-cnn for 3d point cloud classification
  publication-title: 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
– start-page: 28223
  year: 2023
  end-page: 28243
  ident: br0360
  article-title: Contrast with reconstruct: contrastive 3d representation learning guided by generative pretraining
  publication-title: International Conference on Machine Learning
– year: 2024
  ident: br0060
  article-title: Pointwavelet: learning in spectral domain for 3-d point cloud analysis
  publication-title: IEEE Trans. Neural Netw. Learn. Syst.
– volume: 24
  start-page: 1943
  year: 2021
  end-page: 1955
  ident: br0070
  article-title: Geometric back-projection network for point cloud classification
  publication-title: IEEE Trans. Multimed.
– volume: vol. 38
  start-page: 6962
  year: 2024
  end-page: 6970
  ident: br0100
  article-title: Towards compact 3d representations via point feature enhancement masked autoencoders
  publication-title: Proceedings of the AAAI Conference on Artificial Intelligence
– volume: 38
  start-page: 1
  year: 2019
  end-page: 12
  ident: br0030
  article-title: Dynamic graph cnn for learning on point clouds
  publication-title: ACM Trans. Graph.
– start-page: 9224
  year: 2018
  end-page: 9232
  ident: br0260
  article-title: 3d semantic segmentation with submanifold sparse convolutional networks
  publication-title: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition
– year: 2015
  ident: br0380
  article-title: Shapenet: an information-rich 3d model repository
– start-page: 16259
  year: 2021
  end-page: 16268
  ident: br0080
  article-title: Point transformer
  publication-title: Proceedings of the IEEE/CVF International Conference on Computer Vision
– start-page: 626
  year: 2020
  end-page: 642
  ident: br0340
  article-title: Info3d: representation learning on 3d objects using mutual information maximization and contrastive learning
  publication-title: Computer Vision–ECCV 2020: 16th European Conference
– start-page: 1
  year: 2021
  end-page: 11
  ident: br0270
  article-title: Mvtn: multi-view transformation network for 3d shape recognition
  publication-title: Proceedings of the IEEE/CVF International Conference on Computer Vision
– volume: 35
  start-page: 27061
  year: 2022
  ident: 10.1016/j.ins.2025.122474_br0200
  article-title: Point-m2ae: multi-scale masked autoencoders for hierarchical point cloud pre-training
  publication-title: Adv. Neural Inf. Process. Syst.
– ident: 10.1016/j.ins.2025.122474_br0380
– start-page: 9902
  year: 2022
  ident: 10.1016/j.ins.2025.122474_br0330
  article-title: Crosspoint: self-supervised cross-modal contrastive learning for 3d point cloud understanding
– volume: vol. 38
  start-page: 6962
  year: 2024
  ident: 10.1016/j.ins.2025.122474_br0100
  article-title: Towards compact 3d representations via point feature enhancement masked autoencoders
– start-page: 1314
  year: 2019
  ident: 10.1016/j.ins.2025.122474_br0130
  article-title: Searching for mobilenetv3
– start-page: 9267
  year: 2019
  ident: 10.1016/j.ins.2025.122474_br0230
  article-title: Deepgcns: can gcns go as deep as cnns?
– start-page: 1
  year: 2021
  ident: 10.1016/j.ins.2025.122474_br0270
  article-title: Mvtn: multi-view transformation network for 3d shape recognition
– start-page: 19313
  year: 2022
  ident: 10.1016/j.ins.2025.122474_br0280
  article-title: Point-bert: pre-training 3d point cloud transformers with masked point modeling
– volume: 616
  start-page: 526
  year: 2022
  ident: 10.1016/j.ins.2025.122474_br0180
  article-title: Learning user sentiment orientation in social networks for sentiment analysis
  publication-title: Inf. Sci.
  doi: 10.1016/j.ins.2022.10.135
– start-page: 21769
  year: 2023
  ident: 10.1016/j.ins.2025.122474_br0220
  article-title: Learning 3d representations from 2d pre-trained models via image-to-point masked autoencoders
– start-page: 3173
  year: 2021
  ident: 10.1016/j.ins.2025.122474_br0250
  article-title: Paconv: position adaptive convolution with dynamic kernel assembling on point clouds
– ident: 10.1016/j.ins.2025.122474_br0160
– volume: 102
  year: 2024
  ident: 10.1016/j.ins.2025.122474_br0170
  article-title: A cross transformer for image denoising
  publication-title: Inf. Fusion
  doi: 10.1016/j.inffus.2023.102043
– start-page: 6411
  year: 2019
  ident: 10.1016/j.ins.2025.122474_br0240
  article-title: Kpconv: flexible and deformable convolution for point clouds
– volume: 30
  year: 2017
  ident: 10.1016/j.ins.2025.122474_br0020
  article-title: Pointnet++: deep hierarchical feature learning on point sets in a metric space
  publication-title: Adv. Neural Inf. Process. Syst.
– ident: 10.1016/j.ins.2025.122474_br0210
– start-page: 9224
  year: 2018
  ident: 10.1016/j.ins.2025.122474_br0260
  article-title: 3d semantic segmentation with submanifold sparse convolutional networks
– start-page: 1912
  year: 2015
  ident: 10.1016/j.ins.2025.122474_br0400
  article-title: 3d shapenets: a deep representation for volumetric shapes
– start-page: 604
  year: 2022
  ident: 10.1016/j.ins.2025.122474_br0120
  article-title: Masked autoencoders for point cloud self-supervised learning
– volume: 38
  start-page: 1
  year: 2019
  ident: 10.1016/j.ins.2025.122474_br0030
  article-title: Dynamic graph cnn for learning on point clouds
  publication-title: ACM Trans. Graph.
– ident: 10.1016/j.ins.2025.122474_br0370
– start-page: 574
  year: 2020
  ident: 10.1016/j.ins.2025.122474_br0320
  article-title: Pointcontrast: unsupervised pre-training for 3d point cloud understanding
– start-page: 3809
  year: 2021
  ident: 10.1016/j.ins.2025.122474_br0410
  article-title: Revisiting point cloud shape classification with a simple and effective baseline
– start-page: 657
  year: 2022
  ident: 10.1016/j.ins.2025.122474_br0430
  article-title: Masked discrimination for self-supervised learning on point clouds
– start-page: 10296
  year: 2019
  ident: 10.1016/j.ins.2025.122474_br0310
  article-title: Graph attention convolution for point cloud semantic segmentation
– ident: 10.1016/j.ins.2025.122474_br0110
– start-page: 626
  year: 2020
  ident: 10.1016/j.ins.2025.122474_br0340
  article-title: Info3d: representation learning on 3d objects using mutual information maximization and contrastive learning
– start-page: 1534
  year: 2016
  ident: 10.1016/j.ins.2025.122474_br0450
  article-title: 3d semantic parsing of large-scale indoor spaces
– ident: 10.1016/j.ins.2025.122474_br0420
– year: 2024
  ident: 10.1016/j.ins.2025.122474_br0060
  article-title: Pointwavelet: learning in spectral domain for 3-d point cloud analysis
  publication-title: IEEE Trans. Neural Netw. Learn. Syst.
– volume: 24
  start-page: 1943
  year: 2021
  ident: 10.1016/j.ins.2025.122474_br0070
  article-title: Geometric back-projection network for point cloud classification
  publication-title: IEEE Trans. Multimed.
  doi: 10.1109/TMM.2021.3074240
– start-page: 4965
  year: 2021
  ident: 10.1016/j.ins.2025.122474_br0040
  article-title: Adaptive graph convolution for point cloud analysis
– start-page: 652
  year: 2017
  ident: 10.1016/j.ins.2025.122474_br0010
  article-title: Pointnet: deep learning on point sets for 3d classification and segmentation
– volume: 36
  year: 2024
  ident: 10.1016/j.ins.2025.122474_br0440
  article-title: Pointgpt: auto-regressively generative pre-training from point clouds
  publication-title: Adv. Neural Inf. Process. Syst.
– volume: 34
  start-page: 12077
  year: 2021
  ident: 10.1016/j.ins.2025.122474_br0150
  article-title: Segformer: simple and efficient design for semantic segmentation with transformers
  publication-title: Adv. Neural Inf. Process. Syst.
– volume: 23
  start-page: 2019
  year: 2014
  ident: 10.1016/j.ins.2025.122474_br0190
  article-title: Click prediction for web image reranking using multimodal sparse coding
  publication-title: IEEE Trans. Image Process.
  doi: 10.1109/TIP.2014.2311377
– start-page: 28223
  year: 2023
  ident: 10.1016/j.ins.2025.122474_br0360
  article-title: Contrast with reconstruct: contrastive 3d representation learning guided by generative pretraining
– start-page: 746
  year: 2018
  ident: 10.1016/j.ins.2025.122474_br0050
  article-title: Rgcnn: regularized graph cnn for point cloud segmentation
– start-page: 6279
  year: 2018
  ident: 10.1016/j.ins.2025.122474_br0290
  article-title: A graph-cnn for 3d point cloud classification
– volume: 86
  start-page: 42
  year: 2020
  ident: 10.1016/j.ins.2025.122474_br0300
  article-title: Pointngcnn: deep convolutional networks on 3d point clouds with neighborhood graph filters
  publication-title: Comput. Graph.
  doi: 10.1016/j.cag.2019.11.005
– year: 2024
  ident: 10.1016/j.ins.2025.122474_br0350
  article-title: Rethinking masked representation learning for 3d point cloud understanding
  publication-title: IEEE Trans. Image Process.
– volume: 11
  start-page: 231
  year: 2023
  ident: 10.1016/j.ins.2025.122474_br0090
  article-title: Point cloud classification using content-based transformer via clustering in feature space
  publication-title: IEEE/CAA J. Autom. Sin.
  doi: 10.1109/JAS.2023.123432
– start-page: 779
  year: 2016
  ident: 10.1016/j.ins.2025.122474_br0140
  article-title: You only look once: unified, real-time object detection
– start-page: 16259
  year: 2021
  ident: 10.1016/j.ins.2025.122474_br0080
  article-title: Point transformer
– start-page: 1588
  year: 2019
  ident: 10.1016/j.ins.2025.122474_br0390
  article-title: Revisiting point cloud classification: a new benchmark dataset and classification model on real-world data
SSID ssj0004766
Score 2.4763901
Snippet Masked Autoencoders (MAEs) have demonstrated considerable potential in advancing self-supervised learning for 3D point cloud representation. Nevertheless,...
SourceID crossref
elsevier
SourceType Index Database
Publisher
StartPage 122474
SubjectTerms Graph convolution
Graph scale
Masked autoencoder
Point cloud
Title Point-GSMAE: A graph convolution and scale-based masked autoencoder for 3D point cloud representation
URI https://dx.doi.org/10.1016/j.ins.2025.122474
Volume 719
WOSCitedRecordID wos001528665100001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
journalDatabaseRights – providerCode: PRVESC
  databaseName: Elsevier SD Freedom Collection Journals 2021
  issn: 0020-0255
  databaseCode: AIEXJ
  dateStart: 19950101
  customDbUrl:
  isFulltext: true
  dateEnd: 99991231
  titleUrlDefault: https://www.sciencedirect.com
  omitProxy: false
  ssIdentifier: ssj0004766
  providerName: Elsevier
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwtV3dT9RAEN8Q4EEfjKJGVMw8EB8kNW23ZVvfLngKBAmJmBxPzba7DQfaXrieIf71zuxurwtKoia-9C57120z8-t0voex7STlNcIkDLKIo4ESk6MpU2FQJmlZ61DGUpmW-Ufi-DibTPITV3E9N-MERNNk19f57L-yGteQ2VQ6-xfsXm6KC_gdmY5HZDse_4jxJ-206YKPnz-Nxrbs3PSkNunl7romYDBH5uiAXmJq55ucX-KHXHQt9bWk9hKUfchphgNutlN9bRdU5TIbapUaX6t1NU1mb_dKHbzvdt712WKJwjPnot47l-2P8-kyJcj8DyFLvTsGFy0tTqay9d0Tcerq9JY-s75u5kZaJympAVkzvhwWVnb-ItOte-ECDRFqrx6nbykYaEf73GqVTZFnYyQh-EK0zNAqXotFmqPAXhsdjCeHQ8WssFHs_j76eLfJ_Lt1od9rLJ4WcvqQPXDmA4ws2x-xFd1ssPteU8kNtuVKUeA1eHwBJ8QfM-0B5B2MwMADPHgAwgM8eICFB3jwANwX-Hsw8AADD7gJjyfsy4fx6d5-4IZtBBWquB0-plXE5W4ilRRlzaO0UjqvslBleZ3lcVkleVJXaD1HnJRUhc-wEjJSmno2yt2SP2WrTdvoZwxilaqINGv8HRVAUSrNSyWiWMlSyyjbZG96ghYz21Ol6JMNLwqkfkHULyz1N1nSk7xwCLbKXoH4uPu05_922gt2bwDxS7baXS30FluvvnfT-dUrh6Kfim6FXg
linkProvider Elsevier
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Point-GSMAE%3A+A+graph+convolution+and+scale-based+masked+autoencoder+for+3D+point+cloud+representation&rft.jtitle=Information+sciences&rft.au=Bai%2C+Yun&rft.au=Yang%2C+Chaozhi&rft.au=Li%2C+Guanlin&rft.au=He%2C+Xiao&rft.date=2025-11-01&rft.pub=Elsevier+Inc&rft.issn=0020-0255&rft.volume=719&rft_id=info:doi/10.1016%2Fj.ins.2025.122474&rft.externalDocID=S0020025525006061
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=0020-0255&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=0020-0255&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=0020-0255&client=summon