3DViewGraph: Learning Global Features for 3D Shapes from A Graph of Unordered Views with Attention

被引:0
|
作者
Han, Zhizhong [1 ,3 ]
Wang, Xiyang [1 ,2 ]
Chi-Man Vong [4 ]
Liu, Yu-Shen [1 ,2 ]
Zwicker, Matthias [3 ]
Chen, C. L. Philip [5 ]
机构
[1] Tsinghua Univ, Sch Software, Beijing, Peoples R China
[2] Beijing Natl Res Ctr Informat Sci & Technol BNRis, Beijing, Peoples R China
[3] Univ Maryland, Dept Comp Sci, College Pk, MD 20742 USA
[4] Univ Macau, Dept Comp & Informat Sci, Macau, Peoples R China
[5] Univ Macau, Fac Sci & Technol, Macau, Peoples R China
基金
国家重点研发计划;
关键词
NETWORK;
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Learning global features by aggregating information over multiple views has been shown to be effective for 3D shape analysis. For view aggregation in deep learning models, pooling has been applied extensively. However, pooling leads to a loss of the content within views, and the spatial relationship among views, which limits the discriminability of learned features. We propose 3DViewGraph to resolve this issue, which learns 3D global features by more effectively aggregating unordered views with attention. Specifically, unordered views taken around a shape are regarded as view nodes on a view graph. 3DViewGraph first learns a novel latent semantic mapping to project low-level view features into meaningful latent semantic embeddings in a lower dimensional space, which is spanned by latent semantic patterns. Then, the content and spatial information of each pair of view nodes are encoded by a novel spatial pattern correlation, where the correlation is computed among latent semantic patterns. Finally, all spatial pattern correlations are integrated with attention weights learned by a novel attention mechanism. This further increases the discriminability of learned features by highlighting the unordered view nodes with distinctive characteristics and depressing the ones with appearance ambiguity. We show that 3DViewGraph outperforms state-of-the-art methods under three large-scale benchmarks.
引用
收藏
页码:758 / 765
页数:8
相关论文
共 50 条
  • [1] Learning best views of 3D shapes from sketch contour
    Long Zhao
    Shuang Liang
    Jinyuan Jia
    Yichen Wei
    The Visual Computer, 2015, 31 : 765 - 774
  • [2] Learning best views of 3D shapes from sketch contour
    Zhao, Long
    Liang, Shuang
    Jia, Jinyuan
    Wei, Yichen
    VISUAL COMPUTER, 2015, 31 (6-8): : 765 - 774
  • [3] SeqViews2SeqLabels: Learning 3D Global Features via Aggregating Sequential Views by RNN With Attention
    Han, Zhizhong
    Shang, Mingyang
    Liu, Zhenbao
    Vong, Chi-Man
    Liu, Yu-Shen
    Zwicker, Matthias
    Han, Junwei
    Chen, C. L. Philip
    IEEE TRANSACTIONS ON IMAGE PROCESSING, 2019, 28 (02) : 658 - 672
  • [4] An accurate method for 3D object reconstruction from unordered sparse views
    Danilo Djordjevic
    Stevica Cvetković
    Saša V. Nikolić
    Signal, Image and Video Processing, 2017, 11 : 1147 - 1154
  • [5] An accurate method for 3D object reconstruction from unordered sparse views
    Djordjevic, Danilo
    Cvetkovic, Stevica
    Nikolic, Sasa V.
    SIGNAL IMAGE AND VIDEO PROCESSING, 2017, 11 (06) : 1147 - 1154
  • [6] Hierarchical View Predictor: Unsupervised 3D Global Feature Learning through Hierarchical Prediction among Unordered Views
    Han, Zhizhong
    Wang, Xiyang
    Liu, Yu-Shen
    Zwicker, Matthias
    PROCEEDINGS OF THE 29TH ACM INTERNATIONAL CONFERENCE ON MULTIMEDIA, MM 2021, 2021, : 3862 - 3871
  • [7] Inferring 3D Shapes and Deformations from Single Views
    Chen, Yu
    Kim, Tae-Kyun
    Cipolla, Roberto
    COMPUTER VISION-ECCV 2010, PT III, 2010, 6313 : 300 - 313
  • [8] Visual Attention for Rendered 3D Shapes
    Lavoue, Guillaume
    Cordier, Frederic
    Seo, Hyewon
    Larabi, Mohamed-Chaker
    COMPUTER GRAPHICS FORUM, 2018, 37 (02) : 191 - 203
  • [9] 3D2SeqViews: Aggregating Sequential Views for 3D Global Feature Learning by CNN With Hierarchical Attention Aggregation
    Han, Zhizhong
    Lu, Honglei
    Liu, Zhenbao
    Vong, Chi-Man
    Liu, Yu-Shen
    Zwicker, Matthias
    Han, Junwei
    Chen, C. L. Philip
    IEEE TRANSACTIONS ON IMAGE PROCESSING, 2019, 28 (08) : 3986 - 3999
  • [10] Parts4Feature: Learning 3D Global Features from Generally Semantic Parts in Multiple Views
    Han, Zhizhong
    Liu, Xinhai
    Liu, Yu-Shen
    Zwicker, Matthias
    PROCEEDINGS OF THE TWENTY-EIGHTH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE, 2019, : 766 - 773