A study on UAV target detection and 3D positioning methods based on the improved deformable DETR model and multi-view geometry
This paper addresses critical challenges in Unmanned Aerial Vehicle (UAV) target detection and 3D positioning, specifically inaccuracies in localization and lack of robustness in complex environments. The objective of this research is to improve UAV detection and positioning accuracy by proposing an...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
SAGE Publishing
2025-01-01
|
| Series: | Advances in Mechanical Engineering |
| Online Access: | https://doi.org/10.1177/16878132251315505 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832586812733784064 |
|---|---|
| author | Xuebin Liu Hanshan Li |
| author_facet | Xuebin Liu Hanshan Li |
| author_sort | Xuebin Liu |
| collection | DOAJ |
| description | This paper addresses critical challenges in Unmanned Aerial Vehicle (UAV) target detection and 3D positioning, specifically inaccuracies in localization and lack of robustness in complex environments. The objective of this research is to improve UAV detection and positioning accuracy by proposing an enhanced Deformable DETR (Detection Transformer) model integrated with multi-view geometry theory. To achieve this, the study first preprocesses UAV-collected data, then optimizes the convolutional layers of the original DETR model to better handle object occlusion and scale variations. Furthermore, the research incorporates multi-view geometric modeling and multimodal fusion strategies to enhance detection accuracy during the target recognition process. Experimental results demonstrate that the proposed approach achieves over 70% detection accuracy, significantly outperforming traditional methods. The findings underscore the effectiveness of combining the improved Deformable DETR model with multi-view geometry for high-precision detection and 3D localization in complex environments. This research has significant implications for UAV-based applications, such as autonomous navigation, surveillance, and search-and-rescue missions, where precise target detection and 3D positioning are critical for successful operation. |
| format | Article |
| id | doaj-art-b43f6f803fef434792460ce29e3a2b96 |
| institution | Kabale University |
| issn | 1687-8140 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | SAGE Publishing |
| record_format | Article |
| series | Advances in Mechanical Engineering |
| spelling | doaj-art-b43f6f803fef434792460ce29e3a2b962025-01-25T06:03:25ZengSAGE PublishingAdvances in Mechanical Engineering1687-81402025-01-011710.1177/16878132251315505A study on UAV target detection and 3D positioning methods based on the improved deformable DETR model and multi-view geometryXuebin Liu0Hanshan Li1School of Information Engineering, Shanghai Zhongqiao Vocational and Technical University, Shanghai, ChinaSchool of Electronic and Information Engineering, Xi’an Technological University, Xi’an, Shaanxi, ChinaThis paper addresses critical challenges in Unmanned Aerial Vehicle (UAV) target detection and 3D positioning, specifically inaccuracies in localization and lack of robustness in complex environments. The objective of this research is to improve UAV detection and positioning accuracy by proposing an enhanced Deformable DETR (Detection Transformer) model integrated with multi-view geometry theory. To achieve this, the study first preprocesses UAV-collected data, then optimizes the convolutional layers of the original DETR model to better handle object occlusion and scale variations. Furthermore, the research incorporates multi-view geometric modeling and multimodal fusion strategies to enhance detection accuracy during the target recognition process. Experimental results demonstrate that the proposed approach achieves over 70% detection accuracy, significantly outperforming traditional methods. The findings underscore the effectiveness of combining the improved Deformable DETR model with multi-view geometry for high-precision detection and 3D localization in complex environments. This research has significant implications for UAV-based applications, such as autonomous navigation, surveillance, and search-and-rescue missions, where precise target detection and 3D positioning are critical for successful operation.https://doi.org/10.1177/16878132251315505 |
| spellingShingle | Xuebin Liu Hanshan Li A study on UAV target detection and 3D positioning methods based on the improved deformable DETR model and multi-view geometry Advances in Mechanical Engineering |
| title | A study on UAV target detection and 3D positioning methods based on the improved deformable DETR model and multi-view geometry |
| title_full | A study on UAV target detection and 3D positioning methods based on the improved deformable DETR model and multi-view geometry |
| title_fullStr | A study on UAV target detection and 3D positioning methods based on the improved deformable DETR model and multi-view geometry |
| title_full_unstemmed | A study on UAV target detection and 3D positioning methods based on the improved deformable DETR model and multi-view geometry |
| title_short | A study on UAV target detection and 3D positioning methods based on the improved deformable DETR model and multi-view geometry |
| title_sort | study on uav target detection and 3d positioning methods based on the improved deformable detr model and multi view geometry |
| url | https://doi.org/10.1177/16878132251315505 |
| work_keys_str_mv | AT xuebinliu astudyonuavtargetdetectionand3dpositioningmethodsbasedontheimproveddeformabledetrmodelandmultiviewgeometry AT hanshanli astudyonuavtargetdetectionand3dpositioningmethodsbasedontheimproveddeformabledetrmodelandmultiviewgeometry AT xuebinliu studyonuavtargetdetectionand3dpositioningmethodsbasedontheimproveddeformabledetrmodelandmultiviewgeometry AT hanshanli studyonuavtargetdetectionand3dpositioningmethodsbasedontheimproveddeformabledetrmodelandmultiviewgeometry |