3D point cloud object detection algorithm based on Transformer
In response to the difficulty in deploying anchor box based methods in 3D object detection due to the increase in spatial dimensions, this paper studies a point cloud object detection algorithm based on set prediction. This article proposes a Transformer based 3D point cloud object detection algorit...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | zho |
| Published: |
EDP Sciences
2023-12-01
|
| Series: | Xibei Gongye Daxue Xuebao |
| Subjects: | |
| Online Access: | https://www.jnwpu.org/articles/jnwpu/full_html/2023/06/jnwpu2023416p1190/jnwpu2023416p1190.html |
| _version_ | 1827572634087325696 |
|---|---|
| author | LIU Mingyang YANG Qiming HU Guanhua GUO Yan ZHANG Jiandong |
| author_facet | LIU Mingyang YANG Qiming HU Guanhua GUO Yan ZHANG Jiandong |
| author_sort | LIU Mingyang |
| collection | DOAJ |
| description | In response to the difficulty in deploying anchor box based methods in 3D object detection due to the increase in spatial dimensions, this paper studies a point cloud object detection algorithm based on set prediction. This article proposes a Transformer based 3D point cloud object detection algorithm, and combines the characteristics of point clouds in autonomous driving scenarios to propose an improved spatial modulation attention and heat map initialization strategy for training acceleration and query initialization, achieving good detection performance in shallow networks. This article compares it with other algorithms on the KITTI dataset, and the results show that our algorithm has reached an advanced level in performance. We also conducted ablation experiments on the main components of the algorithm to verify the contribution of each module to the detection effect. |
| first_indexed | 2024-03-07T19:07:46Z |
| format | Article |
| id | doaj.art-c2c07fbf1ce84cb5ba922583ea99eab0 |
| institution | Directory Open Access Journal |
| issn | 1000-2758 2609-7125 |
| language | zho |
| last_indexed | 2024-03-07T19:07:46Z |
| publishDate | 2023-12-01 |
| publisher | EDP Sciences |
| record_format | Article |
| series | Xibei Gongye Daxue Xuebao |
| spelling | doaj.art-c2c07fbf1ce84cb5ba922583ea99eab02024-03-01T07:59:43ZzhoEDP SciencesXibei Gongye Daxue Xuebao1000-27582609-71252023-12-014161190119710.1051/jnwpu/20234161190jnwpu2023416p11903D point cloud object detection algorithm based on TransformerLIU Mingyang0YANG Qiming1HU Guanhua2GUO Yan3ZHANG Jiandong4Shenyang Aircraft Design Research InstituteSchool of Electronics and Information, Northwestern Polytechnical UniversitySchool of Electronics and Information, Northwestern Polytechnical UniversityNo. 1 Military Representative Office of Equipment Department of PLA Airforce in ShenyangSchool of Electronics and Information, Northwestern Polytechnical UniversityIn response to the difficulty in deploying anchor box based methods in 3D object detection due to the increase in spatial dimensions, this paper studies a point cloud object detection algorithm based on set prediction. This article proposes a Transformer based 3D point cloud object detection algorithm, and combines the characteristics of point clouds in autonomous driving scenarios to propose an improved spatial modulation attention and heat map initialization strategy for training acceleration and query initialization, achieving good detection performance in shallow networks. This article compares it with other algorithms on the KITTI dataset, and the results show that our algorithm has reached an advanced level in performance. We also conducted ablation experiments on the main components of the algorithm to verify the contribution of each module to the detection effect.https://www.jnwpu.org/articles/jnwpu/full_html/2023/06/jnwpu2023416p1190/jnwpu2023416p1190.htmltransformerspatial modulation attention mechanismheat map initializationtarget detectiondeep learning |
| spellingShingle | LIU Mingyang YANG Qiming HU Guanhua GUO Yan ZHANG Jiandong 3D point cloud object detection algorithm based on Transformer Xibei Gongye Daxue Xuebao transformer spatial modulation attention mechanism heat map initialization target detection deep learning |
| title | 3D point cloud object detection algorithm based on Transformer |
| title_full | 3D point cloud object detection algorithm based on Transformer |
| title_fullStr | 3D point cloud object detection algorithm based on Transformer |
| title_full_unstemmed | 3D point cloud object detection algorithm based on Transformer |
| title_short | 3D point cloud object detection algorithm based on Transformer |
| title_sort | 3d point cloud object detection algorithm based on transformer |
| topic | transformer spatial modulation attention mechanism heat map initialization target detection deep learning |
| url | https://www.jnwpu.org/articles/jnwpu/full_html/2023/06/jnwpu2023416p1190/jnwpu2023416p1190.html |
| work_keys_str_mv | AT liumingyang 3dpointcloudobjectdetectionalgorithmbasedontransformer AT yangqiming 3dpointcloudobjectdetectionalgorithmbasedontransformer AT huguanhua 3dpointcloudobjectdetectionalgorithmbasedontransformer AT guoyan 3dpointcloudobjectdetectionalgorithmbasedontransformer AT zhangjiandong 3dpointcloudobjectdetectionalgorithmbasedontransformer |