Data Preparation Impact on Semantic Segmentation of 3D Mobile LiDAR Point Clouds Using Deep Neural Networks

被引:2
|
作者
Kouhi, Reza Mahmoudi [1 ]
Daniel, Sylvie [1 ]
Giguere, Philippe [2 ]
机构
[1] Univ Laval, Dept Geomat Sci, Quebec City, PQ G1V 0A6, Canada
[2] Univ Laval, Dept Comp Sci & Software Engn, Quebec City, PQ G1V 0A6, Canada
基金
加拿大自然科学与工程研究理事会;
关键词
semantic segmentation; 3D point cloud; deep neural networks; LiDAR;
D O I
10.3390/rs15040982
中图分类号
X [环境科学、安全科学];
学科分类号
08 ; 0830 ;
摘要
Currently, 3D point clouds are being used widely due to their reliability in presenting 3D objects and accurately localizing them. However, raw point clouds are unstructured and do not contain semantic information about the objects. Recently, dedicated deep neural networks have been proposed for the semantic segmentation of 3D point clouds. The focus has been put on the architecture of the network, while the performance of some networks, such as Kernel Point Convolution (KPConv), shows that the way data are presented at the input of the network is also important. Few prior works have studied the impact of using data preparation on the performance of deep neural networks. Therefore, our goal was to address this issue. We propose two novel data preparation methods that are compatible with typical density variations in outdoor 3D LiDAR point clouds. We also investigated two already existing data preparation methods to show their impact on deep neural networks. We compared the four methods with a baseline method based on point cloud partitioning in PointNet++. We experimented with two deep neural networks: PointNet++ and KPConv. The results showed that using any of the proposed data preparation methods improved the performance of both networks by a tangible margin compared to the baseline. The two proposed novel data preparation methods achieved the best results among the investigated methods for both networks. We noticed that, for datasets containing many classes with widely varying sizes, the KNN-based data preparation offered superior performance compared to the Fixed Radius (FR) method. Moreover, this research allowed identifying guidelines to select meaningful downsampling and partitioning of large-scale outdoor 3D LiDAR point clouds at the input of deep neural networks.
引用
收藏
页数:19
相关论文
共 50 条
  • [1] Bilevel Convolutional Neural Networks for 3D Semantic Segmentation Using Large-scale LiDAR Point Clouds in Complex Environments
    Jiang, Tengping
    Yang, Bisheng
    Zhou, Yuzhou
    Zhu, Runsong
    Hu, Zongtian
    Dong, Zhen
    [J]. Wuhan Daxue Xuebao (Xinxi Kexue Ban)/Geomatics and Information Science of Wuhan University, 2020, 45 (12): : 1942 - 1948
  • [2] On the Segmentation of 3D LIDAR Point Clouds
    Douillard, B.
    Underwood, J.
    Kuntz, N.
    Vlaskine, V.
    Quadros, A.
    Morton, P.
    Frenkel, A.
    [J]. 2011 IEEE INTERNATIONAL CONFERENCE ON ROBOTICS AND AUTOMATION (ICRA), 2011,
  • [3] Domain Transfer for Semantic Segmentation of LiDAR Data using Deep Neural Networks
    Langer, Ferdinand
    Milioto, Andres
    Haag, Alexandre
    Behley, Jens
    Stachniss, Cyrill
    [J]. 2020 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS), 2020, : 8263 - 8270
  • [4] Automated semantic segmentation of 3D point clouds of railway tunnel using deep learning
    Park, Jeongjun
    Kim, Byung-Kyu
    Lee, Jun S.
    Yoo, Mintaek
    Lee, Il-Wha
    Ryu, Young-Moo
    [J]. PROCEEDINGS OF THE ITA-AITES WORLD TUNNEL CONGRESS 2023, WTC 2023: Expanding Underground-Knowledge and Passion to Make a Positive Impact on the World, 2023, : 2844 - 2852
  • [5] SEMANTIC SEGMENTATION OF INDOOR 3D POINT CLOUDS BY JOINT OPTIMIZATION OF GEOMETRIC FEATURES AND NEURAL NETWORKS
    Yao, M. M.
    Li, X. M.
    Wang, W. X.
    Xie, L. F.
    Tang, S. J.
    [J]. 17TH 3D GEOINFO CONFERENCE, 2022, 10-4 (W2): : 305 - 310
  • [6] SEGCloud: Semantic Segmentation of 3D Point Clouds
    Tchapmi, Lyne P.
    Choy, Christopher B.
    Armeni, Iro
    Gwak, JunYoung
    Savarese, Silvio
    [J]. PROCEEDINGS 2017 INTERNATIONAL CONFERENCE ON 3D VISION (3DV), 2017, : 537 - 547
  • [7] AUTOMATED TRAINING DATA CREATION FOR SEMANTIC SEGMENTATION OF 3D POINT CLOUDS
    De Geyter, Sam
    Bassier, Maarten
    Vergauwen, Maarten
    [J]. MEASUREMENT, VISUALISATION AND PROCESSING IN BIM FOR DESIGN AND CONSTRUCTION MANAGEMENT II, 2022, 46-5 (W1): : 59 - 67
  • [8] DPRNet: Deep 3D Point Based Residual Network for Semantic Segmentation and Classification of 3D Point Clouds
    Arshad, Saira
    Shahzad, Muhammad
    Riaz, Qaiser
    Fraz, Muhammad Moazam
    [J]. IEEE ACCESS, 2019, 7 : 68892 - 68904
  • [9] Head pose estimation using deep neural networks and 3D point clouds
    Xu, Yuanquan
    Jung, Cheolkon
    Chang, Yakun
    [J]. PATTERN RECOGNITION, 2022, 121
  • [10] A Simple Deep Learning Network for Classification of 3D Mobile LiDAR Point Clouds
    Yanjun WANG
    Shaochun LI
    Mengjie WANG
    Yunhao LIN
    [J]. Journal of Geodesy and Geoinformation Science, 2021, 4 (03) : 49 - 59