摘要
With the increasing popularity of 3D sensors(e.g.,Kinect)and light field cameras,technologies such as driverless,smart home and virtual reality have become hot spots for engineering applications.As an important part of 3D vision tasks,point cloud semantic segmentation has received a lot of attention from researchers.In this work,we focus on realistically collected indoor point cloud data and propose a point cloud semantic segmentation method based on PAConv and SE_variant.The SE_variant module captures global perception from a broad perspective of feature space by fusing different pooling methods,which fully utilize the channel information of point clouds.The effectiveness of the method is verified by comparing with other methods on S3DIS and ScanNetV2 semantic tagging benchmarks,and achieving 65.3%mIoU in S3DIS,47.6%mIoU in ScanNetV2.The results of the ablation experiments verify the effectiveness of the key modules and analyze how to use the attention mechanism to improve the 3D semantic segmentation performance.
作者简介
ZHANG Ying received the Ph.D.degree from Hunan University in 2010.He is now a lecfurer jn xiangtan University.His research interests include the development of electric vehicles controller and automotive Internet applications.E-mail:zhangying@xtu.edu.cn;SUN Yue received the B.E.degree from Hunan University of Technology and Business in 2020.She is now a M.E.candidate in Xiangtan University.Her main research interests include pattern recognition and image processing.E-mail:sunyue1049@163.com;WU Lin received the B.E.degree from Xiangtan University in 2021.She is now a M.E.candidate in Xiangtan University.Her main research interests include pattern recognition and image processing.E-mail:wl19060@163.com;ZHANG Lulu received the B.Sc.degree from University of Science and Technology Liaoning in 2020.She is now a M.E.candidate in Xiangtan University.Her main research interests include pattern recognition and image processing.E-mail:208366092@qq.com;MENG Bumin received the Ph.D.from Hunan University in 2018.He is now a lecturer in Xiangtan University.His research interests include the development of electric vehicles controller and automotive Internet applications.E-mail:mengbm@163.com
