YOLOv7-3D: A Monocular 3D Traffic Object Detection Method from a Roadside Perspective
Current autonomous driving systems predominantly focus on 3D object perception from the vehicle’s perspective. However, the single-camera 3D object detection algorithm in the roadside monitoring scenario provides stereo perception of traffic objects, offering more accurate collection and analysis of...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-10-01
|
| Series: | Applied Sciences |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-3417/13/20/11402 |
| _version_ | 1797574916459986944 |
|---|---|
| author | Zixun Ye Hongying Zhang Jingliang Gu Xue Li |
| author_facet | Zixun Ye Hongying Zhang Jingliang Gu Xue Li |
| author_sort | Zixun Ye |
| collection | DOAJ |
| description | Current autonomous driving systems predominantly focus on 3D object perception from the vehicle’s perspective. However, the single-camera 3D object detection algorithm in the roadside monitoring scenario provides stereo perception of traffic objects, offering more accurate collection and analysis of traffic information to ensure reliable support for urban traffic safety. In this paper, we propose the YOLOv7-3D algorithm specifically designed for single-camera 3D object detection from a roadside viewpoint. Our approach utilizes various information, including 2D bounding boxes, projected corner keypoints, and offset vectors relative to the center of the 2D bounding boxes, to enhance the accuracy of 3D object bounding box detection. Additionally, we introduce a 5-layer feature pyramid network (FPN) structure and a multi-scale spatial attention mechanism to improve feature saliency for objects of different scales, thereby enhancing the detection accuracy of the network. Experimental results demonstrate that our YOLOv7-3D network achieved significantly higher detection accuracy on the Rope3D dataset while reducing computational complexity by 60%. |
| first_indexed | 2024-03-10T21:28:52Z |
| format | Article |
| id | doaj.art-e1f935170e424b9da8848cb0a67b37f2 |
| institution | Directory Open Access Journal |
| issn | 2076-3417 |
| language | English |
| last_indexed | 2024-03-10T21:28:52Z |
| publishDate | 2023-10-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Applied Sciences |
| spelling | doaj.art-e1f935170e424b9da8848cb0a67b37f22023-11-19T15:31:43ZengMDPI AGApplied Sciences2076-34172023-10-0113201140210.3390/app132011402YOLOv7-3D: A Monocular 3D Traffic Object Detection Method from a Roadside PerspectiveZixun Ye0Hongying Zhang1Jingliang Gu2Xue Li3School of Information Engineering, Southwest University of Science and Technology, Mianyang 621010, ChinaSchool of Information Engineering, Southwest University of Science and Technology, Mianyang 621010, ChinaInstitute of Applied Electronics, CAEP, Mianyang 621900, ChinaSchool of Information Engineering, Southwest University of Science and Technology, Mianyang 621010, ChinaCurrent autonomous driving systems predominantly focus on 3D object perception from the vehicle’s perspective. However, the single-camera 3D object detection algorithm in the roadside monitoring scenario provides stereo perception of traffic objects, offering more accurate collection and analysis of traffic information to ensure reliable support for urban traffic safety. In this paper, we propose the YOLOv7-3D algorithm specifically designed for single-camera 3D object detection from a roadside viewpoint. Our approach utilizes various information, including 2D bounding boxes, projected corner keypoints, and offset vectors relative to the center of the 2D bounding boxes, to enhance the accuracy of 3D object bounding box detection. Additionally, we introduce a 5-layer feature pyramid network (FPN) structure and a multi-scale spatial attention mechanism to improve feature saliency for objects of different scales, thereby enhancing the detection accuracy of the network. Experimental results demonstrate that our YOLOv7-3D network achieved significantly higher detection accuracy on the Rope3D dataset while reducing computational complexity by 60%.https://www.mdpi.com/2076-3417/13/20/11402object detectionmonocular 3D object detectionroadside perspective |
| spellingShingle | Zixun Ye Hongying Zhang Jingliang Gu Xue Li YOLOv7-3D: A Monocular 3D Traffic Object Detection Method from a Roadside Perspective Applied Sciences object detection monocular 3D object detection roadside perspective |
| title | YOLOv7-3D: A Monocular 3D Traffic Object Detection Method from a Roadside Perspective |
| title_full | YOLOv7-3D: A Monocular 3D Traffic Object Detection Method from a Roadside Perspective |
| title_fullStr | YOLOv7-3D: A Monocular 3D Traffic Object Detection Method from a Roadside Perspective |
| title_full_unstemmed | YOLOv7-3D: A Monocular 3D Traffic Object Detection Method from a Roadside Perspective |
| title_short | YOLOv7-3D: A Monocular 3D Traffic Object Detection Method from a Roadside Perspective |
| title_sort | yolov7 3d a monocular 3d traffic object detection method from a roadside perspective |
| topic | object detection monocular 3D object detection roadside perspective |
| url | https://www.mdpi.com/2076-3417/13/20/11402 |
| work_keys_str_mv | AT zixunye yolov73damonocular3dtrafficobjectdetectionmethodfromaroadsideperspective AT hongyingzhang yolov73damonocular3dtrafficobjectdetectionmethodfromaroadsideperspective AT jinglianggu yolov73damonocular3dtrafficobjectdetectionmethodfromaroadsideperspective AT xueli yolov73damonocular3dtrafficobjectdetectionmethodfromaroadsideperspective |