LSPConv: local spatial projection convolution for point cloud analysis
This study introduces a novel approach, Local Spatial Projection Convolution (LSPConv), for point cloud classification and semantic segmentation. Unlike conventional methods utilizing relative coordinates for local geometric information, our motivation stems from the inadequacy of existing technique...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
PeerJ Inc.
2024-01-01
|
| Series: | PeerJ Computer Science |
| Subjects: | |
| Online Access: | https://peerj.com/articles/cs-1738.pdf |
| _version_ | 1827388031945932800 |
|---|---|
| author | Haoming Zhang Ke Wang Chen Zhong Kaijie Yun Zilong Wang Yifan Yang Xianshui Tao |
| author_facet | Haoming Zhang Ke Wang Chen Zhong Kaijie Yun Zilong Wang Yifan Yang Xianshui Tao |
| author_sort | Haoming Zhang |
| collection | DOAJ |
| description | This study introduces a novel approach, Local Spatial Projection Convolution (LSPConv), for point cloud classification and semantic segmentation. Unlike conventional methods utilizing relative coordinates for local geometric information, our motivation stems from the inadequacy of existing techniques for representing the intricate spatial organization of unconsolidated and irregular 3D point clouds. To address this limitation, we propose a Local Spatial Projection Module utilizing a vector projection strategy, designed to capture comprehensive local spatial information more effectively. Moreover, recent studies emphasize the importance of anisotropic kernels for point cloud feature extraction, considering the distinct contributions of individual neighboring points. To cater to this requirement, we introduce the Feature Weight Assignment (FWA) Module to assign weights to neighboring points, enhancing the anisotropy crucial for accurate feature extraction. Additionally, we introduce an Anisotropic Relative Feature Encoding Module that adaptively encodes points based on their relative features, further amplifying the anisotropic characteristics. Our approaches achieve remarkable results for point cloud classification and segmentation in several benchmark datasets based on extensive qualitative and quantitative evaluation. |
| first_indexed | 2024-03-08T16:11:52Z |
| format | Article |
| id | doaj.art-e25ed35cc9e341aa97ff82d07e4442c3 |
| institution | Directory Open Access Journal |
| issn | 2376-5992 |
| language | English |
| last_indexed | 2024-03-08T16:11:52Z |
| publishDate | 2024-01-01 |
| publisher | PeerJ Inc. |
| record_format | Article |
| series | PeerJ Computer Science |
| spelling | doaj.art-e25ed35cc9e341aa97ff82d07e4442c32024-01-07T15:05:09ZengPeerJ Inc.PeerJ Computer Science2376-59922024-01-0110e173810.7717/peerj-cs.1738LSPConv: local spatial projection convolution for point cloud analysisHaoming Zhang0Ke Wang1Chen Zhong2Kaijie Yun3Zilong Wang4Yifan Yang5Xianshui Tao6State Key Laboratory of Robotics and Systems, Harbin Institute of Technology, Harbin, ChinaState Key Laboratory of Robotics and Systems, Harbin Institute of Technology, Harbin, ChinaShenyang Fire Science and Technology Research Institute of MEM, Shenyang, ChinaState Key Laboratory of Robotics and Systems, Harbin Institute of Technology, Harbin, ChinaState Key Laboratory of Robotics and Systems, Harbin Institute of Technology, Harbin, ChinaState Key Laboratory of Robotics and Systems, Harbin Institute of Technology, Harbin, ChinaWuhu Hit Robot Technology Research Institute Co., Ltd., Wuhu, Anhui, ChinaThis study introduces a novel approach, Local Spatial Projection Convolution (LSPConv), for point cloud classification and semantic segmentation. Unlike conventional methods utilizing relative coordinates for local geometric information, our motivation stems from the inadequacy of existing techniques for representing the intricate spatial organization of unconsolidated and irregular 3D point clouds. To address this limitation, we propose a Local Spatial Projection Module utilizing a vector projection strategy, designed to capture comprehensive local spatial information more effectively. Moreover, recent studies emphasize the importance of anisotropic kernels for point cloud feature extraction, considering the distinct contributions of individual neighboring points. To cater to this requirement, we introduce the Feature Weight Assignment (FWA) Module to assign weights to neighboring points, enhancing the anisotropy crucial for accurate feature extraction. Additionally, we introduce an Anisotropic Relative Feature Encoding Module that adaptively encodes points based on their relative features, further amplifying the anisotropic characteristics. Our approaches achieve remarkable results for point cloud classification and segmentation in several benchmark datasets based on extensive qualitative and quantitative evaluation.https://peerj.com/articles/cs-1738.pdfDeep learningPoint cloudSemantic segmentationClassificationAnisotropic kernelWeight assignment |
| spellingShingle | Haoming Zhang Ke Wang Chen Zhong Kaijie Yun Zilong Wang Yifan Yang Xianshui Tao LSPConv: local spatial projection convolution for point cloud analysis PeerJ Computer Science Deep learning Point cloud Semantic segmentation Classification Anisotropic kernel Weight assignment |
| title | LSPConv: local spatial projection convolution for point cloud analysis |
| title_full | LSPConv: local spatial projection convolution for point cloud analysis |
| title_fullStr | LSPConv: local spatial projection convolution for point cloud analysis |
| title_full_unstemmed | LSPConv: local spatial projection convolution for point cloud analysis |
| title_short | LSPConv: local spatial projection convolution for point cloud analysis |
| title_sort | lspconv local spatial projection convolution for point cloud analysis |
| topic | Deep learning Point cloud Semantic segmentation Classification Anisotropic kernel Weight assignment |
| url | https://peerj.com/articles/cs-1738.pdf |
| work_keys_str_mv | AT haomingzhang lspconvlocalspatialprojectionconvolutionforpointcloudanalysis AT kewang lspconvlocalspatialprojectionconvolutionforpointcloudanalysis AT chenzhong lspconvlocalspatialprojectionconvolutionforpointcloudanalysis AT kaijieyun lspconvlocalspatialprojectionconvolutionforpointcloudanalysis AT zilongwang lspconvlocalspatialprojectionconvolutionforpointcloudanalysis AT yifanyang lspconvlocalspatialprojectionconvolutionforpointcloudanalysis AT xianshuitao lspconvlocalspatialprojectionconvolutionforpointcloudanalysis |