一种探讨点云深度学习决策的PointNet++解析网络A PointNet++ Analytic Network that Explores Point Cloud Deep Learning Decision-Making
龚国栋;李耀斌;花向红;赵不钒;卢荣;
摘要(Abstract)：
针对三维点云数据分类深度学习可解释性研究，提出一种探讨点云深度学习决策的PointNet++解析网络，探索隐藏在PointNet++网络中的特征信息。根据二维图像解译工作中的类激活映射图，提出了三维点云的类激活映射图，并将点云类激活映射图作为探索PointNet++网络分类决策的依据，采用多层感知机取代全连接层，并使用均值池化层来聚合卷积特征。实验数据为ModelNet40数据集，验证了所提出的PointNet++解析网络的可行性。研究结果表明，所提算法达到了较高的分类精度并且能够对PointNet++分类决策进行探讨，提取直接有助于决策制定的特征区域。
关键词(KeyWords)： 点云;深度学习;PointNet++;解析网络
基金项目(Foundation): 国家自然科学基金（41674005,41871373）
作者(Authors): 龚国栋;李耀斌;花向红;赵不钒;卢荣;
DOI: 10.14188/j.2095-6045.2020164
参考文献(References)：
- [1]俞斌，董晨，刘延华，等.基于深度学习的点云分割方法综述[J].计算机工程与应用，2020,56(1):38-45
- [2]赵中阳，程英蕾，释小松，等.基于多尺度特征和PointNet的LiDAR点云地物分类方法[J].激光与光电子学进展，2019,56(5):251-258
- [3]牛辰庚，刘玉杰，李宗民，等.基于点云数据的三维目标识别和模型分割方法[J].图学学报，2019,40(2):274-281
- [4]白静，司庆龙，秦飞巍.轻量级实时点云分类网络LightPointNet[J].计算机辅助设计与图形学学报，2019,31(4):612-621
- [5] Qi C R,Yi L,Su H,et al. PointNet++:Deep Hierarchical Feature Learning on Point Sets in a Metric Space[J]. Neural Information Processing Systems,2017:30-38
- [6] Zhang Q S,Cao R M,Wu Y N,et al. Mining Object Parts from CNNs via Active Question-Answering[C].IEEE Conference on Computer Vision and Pattern Recognition(CVPR),California,Los Angeles,USA,2017
- [7] Zeiler M D,Rob F. Visualizing and Understanding Convolutional Networks[C]. European Conference on Computer Vision,Zurich,Switzerland,2014
- [8] Zhang Q S,Wang W G,Zhu S C. Examining CNN Representations with Respect to Dataset Bias[C].Thirty-One AAAI Conference on Artificial Intelligence,San Francisco,USA,2017
- [9] Wang Y,Sun Y B,Liu Z W,et al. Dynamic Graph CNN for Learning on Point Clouds[J]. Acm Transactions on Graphics,arXiv:1801. 07829,2018
- [10] Shen Y,Feng C,Yang Y,et al. Mining Point Cloud Local Structures by Kernel Correlation and Graph Pooling[J]. Computer Vision and Pattern Recognition,arXiv:1712. 06760,2017
- [11] Charles R Q,Su H,Kaichun M,et al. PointNet:Deep Learning on Point Sets for 3D Classification and Segmentation[C]. 2017 IEEE Conference on Computer Vision and Pattern Recognition(CVPR). Honolulu,HI,USA,2017
- [12] Zhang Quanshi,Cao Ruiming,Shi Feng,et al. Interpreting CNN Knowledge via an Explanatory Graph[C]. Thirty-Second AAAI Conference on Artificial Intelligence,New Orleans,USA,2018
- [13] Zhou B,Khosla A,Lapedriza A,et al. Learning Deep Features for Discriminative Localization[J]. Multimedia,arXiv:1810. 06030,2015