An efficient image-guided-based 3D point cloud moving object segmentation with transformer-attention in autonomous driving
For intelligent transportation systems, moving object segmentation (MOS) provides valuable information for robots and intelligent vehicles, such as collision avoidance, path planning, and static map construction. However, all existing 3D point cloud MOS methods are based on LiDAR-only, which limits...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2023-09-01
|
| Series: | International Journal of Applied Earth Observations and Geoinformation |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S1569843223003126 |
| _version_ | 1827810099373015040 |
|---|---|
| author | Qipeng Li Yuan Zhuang |
| author_facet | Qipeng Li Yuan Zhuang |
| author_sort | Qipeng Li |
| collection | DOAJ |
| description | For intelligent transportation systems, moving object segmentation (MOS) provides valuable information for robots and intelligent vehicles, such as collision avoidance, path planning, and static map construction. However, all existing 3D point cloud MOS methods are based on LiDAR-only, which limits the ability to fuse supplementary information from different sensors. In this article, we solve the robust and accurate 3D MOS problem by designing a dual-stream network that integrates point clouds and images. We propose a perspective residual mechanism to mine the spatio-temporal motion information of point clouds, and design a fusion module based on Transformer Attention to extract multi-scale feature information from point clouds and images, improving the segmentation integrity of moving objects. Many experiments on the benchmark dataset show the superiority of our method. On the Semantic-KITTI, we outperform the advanced method by 6.5% mIoU. And we further apply our proposed model to the Semantic-KITTI: Moving Object Segmentation competition and achieve an advanced ranking on the leaderboard. |
| first_indexed | 2024-03-11T22:48:58Z |
| format | Article |
| id | doaj.art-a372b0fb31184e03a9a6859a87f130e5 |
| institution | Directory Open Access Journal |
| issn | 1569-8432 |
| language | English |
| last_indexed | 2024-03-11T22:48:58Z |
| publishDate | 2023-09-01 |
| publisher | Elsevier |
| record_format | Article |
| series | International Journal of Applied Earth Observations and Geoinformation |
| spelling | doaj.art-a372b0fb31184e03a9a6859a87f130e52023-09-22T04:38:22ZengElsevierInternational Journal of Applied Earth Observations and Geoinformation1569-84322023-09-01123103488An efficient image-guided-based 3D point cloud moving object segmentation with transformer-attention in autonomous drivingQipeng Li0Yuan Zhuang1State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan 430079, ChinaCorresponding author.; State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan 430079, ChinaFor intelligent transportation systems, moving object segmentation (MOS) provides valuable information for robots and intelligent vehicles, such as collision avoidance, path planning, and static map construction. However, all existing 3D point cloud MOS methods are based on LiDAR-only, which limits the ability to fuse supplementary information from different sensors. In this article, we solve the robust and accurate 3D MOS problem by designing a dual-stream network that integrates point clouds and images. We propose a perspective residual mechanism to mine the spatio-temporal motion information of point clouds, and design a fusion module based on Transformer Attention to extract multi-scale feature information from point clouds and images, improving the segmentation integrity of moving objects. Many experiments on the benchmark dataset show the superiority of our method. On the Semantic-KITTI, we outperform the advanced method by 6.5% mIoU. And we further apply our proposed model to the Semantic-KITTI: Moving Object Segmentation competition and achieve an advanced ranking on the leaderboard.http://www.sciencedirect.com/science/article/pii/S1569843223003126Moving object segmentationMulti-sensor fusionTransformerIntelligent transportation system |
| spellingShingle | Qipeng Li Yuan Zhuang An efficient image-guided-based 3D point cloud moving object segmentation with transformer-attention in autonomous driving International Journal of Applied Earth Observations and Geoinformation Moving object segmentation Multi-sensor fusion Transformer Intelligent transportation system |
| title | An efficient image-guided-based 3D point cloud moving object segmentation with transformer-attention in autonomous driving |
| title_full | An efficient image-guided-based 3D point cloud moving object segmentation with transformer-attention in autonomous driving |
| title_fullStr | An efficient image-guided-based 3D point cloud moving object segmentation with transformer-attention in autonomous driving |
| title_full_unstemmed | An efficient image-guided-based 3D point cloud moving object segmentation with transformer-attention in autonomous driving |
| title_short | An efficient image-guided-based 3D point cloud moving object segmentation with transformer-attention in autonomous driving |
| title_sort | efficient image guided based 3d point cloud moving object segmentation with transformer attention in autonomous driving |
| topic | Moving object segmentation Multi-sensor fusion Transformer Intelligent transportation system |
| url | http://www.sciencedirect.com/science/article/pii/S1569843223003126 |
| work_keys_str_mv | AT qipengli anefficientimageguidedbased3dpointcloudmovingobjectsegmentationwithtransformerattentioninautonomousdriving AT yuanzhuang anefficientimageguidedbased3dpointcloudmovingobjectsegmentationwithtransformerattentioninautonomousdriving AT qipengli efficientimageguidedbased3dpointcloudmovingobjectsegmentationwithtransformerattentioninautonomousdriving AT yuanzhuang efficientimageguidedbased3dpointcloudmovingobjectsegmentationwithtransformerattentioninautonomousdriving |