无人机高空航拍视角下小尺度车辆精确检测方法

doi:10.16097/j.cnki.1009-6744.2024.03.029

交通运输系统工程与信息 ›› 2024, Vol. 24 ›› Issue (3): 299-309.DOI: 10.16097/j.cnki.1009-6744.2024.03.029

无人机高空航拍视角下小尺度车辆精确检测方法

张河山^{a, b}，谭鑫^a，范梦伟^a，潘存书^a，徐进^{a, b}，张羽^*c

重庆交通大学，a. 交通运输学院；b. 山区复杂道路环境“人-车-路”协同与安全重庆市重点实验室；c. 经济管理学院，重庆 400074

收稿日期:2024-02-04 修回日期:2024-03-05 接受日期:2024-03-19 出版日期:2024-06-25 发布日期:2024-06-24
作者简介:张河山(1988- )，男，重庆万州人，讲师，博士
基金资助:
国家自然科学基金(52172340)；重庆市教育委员会青年项目(KJQN202200710)；重庆市博士后科学基金(CSTB2022NSCQ-BHX0731)

Accurate Detection Method of Small-scale Vehicles from Perspective of Unmanned Aerial Vehicle High-altitude Aerial Photography

ZHANG Heshan^{a, b}, TAN Xin^a , FAN Mengwei^a , PAN Cunshu^a , XU Jin^{a, b}, ZHANG Yu^*c

a. School of Traffic & Transportation; b. Chongqing Key Laboratory of "Human-Vehicle-Road" Cooperation and Safety for Mountain Complex Environment; c. School of Economics and Management, Chongqing Jiaotong University, Chongqing 400074, China

Received:2024-02-04 Revised:2024-03-05 Accepted:2024-03-19 Online:2024-06-25 Published:2024-06-24
Supported by:
National Natural Science Foundation of China (52172340)；Chongqing Education Committee Youth Project (KJQN202200710)；Chongqing Postdoctoral Science Foundation (CSTB2022NSCQ-BHX0731)

摘要/Abstract

摘要： 无人机高空航拍图像中车辆像素占比极低，目标可视化信息较少，在目标检测任务中容易漏检和误检。因此，本文提出一种基于改进YOLOX(You Only Look Once X)的无人机高空航拍视角下小尺度车辆精确检测方法。首先，为增强网络对低级特征的提取能力，在原始YOLOX预测头部增加一个160 pixel×160 pixel的浅层特征提取网络；其次，在骨干网络后端嵌入基于归一化的注意力机制模块(Normalization-based Attention Module, NAM)，以抑制冗余的非显著特征表达；最后，为了增大小尺度车辆的相对像素比，提升网络捕捉有效特征信息的能力，提出一种基于滑动窗口的图像切分检测方法。试验结果表明，改进YOLOX网络表现出良好的检测效能，检测精度达到了84.58%，优于典型的目标检测网络Faster R-CNN(79.95%)、YOLOv3(83.69%)、YOLOv5(84.31%)及YOLOX(83.10%)。此外，改进YOLOX能够有效解决无人机高空航拍图像中小尺度车辆的漏检和误检问题，且预测框更贴合车辆的实际轮廓；同时，在不同航拍高度的目标检测任务中具有较高的鲁棒性。

关键词: 智能交通, 小尺度车辆检测, YOLOX, 无人机, 注意力机制, 浅层特征提取网络

Abstract: The proportion of vehicle pixels in high-altitude aerial images taken from unmanned aerial vehicle (UAV) is low, and there are limited visual information of can be identified for the targets, which result in missed or false detection in the detection tasks. This paper proposes a detection method of small-scale vehicles based on improved YOLOX (You Only Look Once X) from the perspective of high-altitude aerial photography. First, to enhance the network 's ability to extract low-level features, this study added a shallow feature extraction network of 160 pixel× 160 pixel to the original YOLOX prediction head. Then, a Normalization-based Attention Module (NAM) was embedded after the backbone network to suppress redundant non-significant feature expression. To increase the relative pixel ratio of small-scale vehicles and improve the ability of the network to capture effective feature information, an image segmentation detection method was proposed based on sliding window. The experiment results show that the improved YOLOX network shows good detection performance, and the detection accuracy reaches 84.58%, which is better than the typical target detection network Faster R-CNN (79.95%), YOLOv3 (83.69%), YOLOv5 (84.31% ), and YOLOX (83.10% ). In addition, the improved YOLOX can effectively solve the problem of missed detection and false detection of small-scale vehicles in high-altitude aerial images of UAV, and the prediction box is more suitable for the actual contour of the vehicle. At the same time, it has high robustness in target detection tasks at different aerial heights.

Key words: intelligent transportation, small-scale vehicle detection, YOLOX, UAV, attention mechanism, shallow feature extraction network

中图分类号:

U491

张河山, 谭鑫, 范梦伟, 潘存书, 徐进, 张羽. 无人机高空航拍视角下小尺度车辆精确检测方法[J]. 交通运输系统工程与信息, 2024, 24(3): 299-309.

ZHANG Heshan, TAN Xin , FAN Mengwei, PAN Cunshu, XU Jin, ZHANG Yu. Accurate Detection Method of Small-scale Vehicles from Perspective of Unmanned Aerial Vehicle High-altitude Aerial Photography[J]. Journal of Transportation Systems Engineering and Information Technology, 2024, 24(3): 299-309.

导出引用管理器 EndNote|Ris|BibTeX

链接本文: http://www.tseit.org.cn/CN/10.16097/j.cnki.1009-6744.2024.03.029

http://www.tseit.org.cn/CN/Y2024/V24/I3/299

参考文献

[1] 陈志军, 胡军楠, 冷姚, 等. 基于轻量化网络和注意力机制的智能车快速目标识别方法[J]. 交通运输系统工程与信息, 2022, 22(6): 105-113. [CHEN Z J, HU J N, LENG Y, et al. Intelligent vehicle target fast recognition based on lightweight network and attention mechanism [J]. Journal of Transportation Systems Engineering and Information Technology, 2022, 22(6): 105-113.]

[2] REN S, HE K, GIRSHICK R, et al. Faster R-CNN: Towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6): 1137-1149.

[3] REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: Unified, real-time object detection[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas: IEEE, 2016: 779– 788.

[4] GE Z, LIU S, WANG F, et al. YOLOX: Exceeding YOLO series in 2021 [EB/OL]. (2021-08-06) [2023-12-30]. https://arxiv.org/abs/2107.08430.

[5] 徐光达, 毛国君. 多层级特征融合的无人机航拍图像目标检测[J]. 计算机科学与探索, 2023, 17(3): 635- 645. [XU G D, MAO G J. Aerial image object detection of UAV based on multi-level feature fusion[J]. Journal of Frontiers of Computer Science and Technology, 2023, 17 (3): 635-645.]

[6] 彭博, 蔡晓禹, 唐聚, 等. 基于形态检测与深度学习的高空视频车辆识别[J]. 交通运输系统工程与信息, 2019, 19(6): 45-51. [PENG B, CAI X Y, TANG J, et al. Morphous detection and deep learning based approach of vehicle recognition in aerial videos[J]. Journal of Transportation Systems Engineering and Information Technology, 2019, 19(6): 45-51.]

[7] 张河山, 范梦伟, 谭鑫, 等. 基于改进YOLOX的无人机航拍图像密集小目标车辆检测[J/OL]. 吉林大学学报(工学版), (2023-12-28) [2023-12-30]. https://doi.org/ 10.13229/j.cnki.jdxbgxb.20230779. [ZHANG H S, FANG M W, TAN X, et al. Dense small object vehicle detection in UAV aerial images using improved YOLOX [J/OL]. Journal of Jilin University(Engineering and Technology Edition), (2023-12-28) [2023-12-30]. https://doi.org/10.13229/j.cnki.jdxbgxb.20230779.]

[8] 易华辉, 宋文治, 黄金香, 等. 基于改进YOLOv5的无人机小目标检测方法研究[J]. 机电工程技术, 2023, 52 (2): 139-144. [YI H H, SONG W Z, HUANG J X, et al. UAV small target detection based on improved YOLOv5 [J]. Mechanical & Electrical Engineering Technology, 2023, 52(2): 139-144.]

[9] ZHANG H S, TAN X, FAN M W, et al. Accurate detection and tracking of small-scale vehicles in highaltitude unmanned aerial vehicle bird-view imagery [J/OL]. Journal of Advanced Transportation, (2023- 12-23) [2023-12-30]. https://doi.org/10.1155/2023/ 5384844.

[10] LIN Q, DING Y, XU H, et al. Ecascade-rcnn: Enhanced cascade rcnn for multi-scale object detection in uav images[C]//2021 7th International Conference on Automation, Robotics and Applications(ICARA), IEEE, 2021: 268-272.

[11] AMUDHAN A N, SUDHEER A P. Lightweight and computationally faster hypermetropic convolutional neural network for small size object detection[J]. Image and Vision Computing, 2022, 119: 104396

[12] LIU Y C, SHAO Z R, TENG Y Y, et al. NAM: Normalization-based attention module [EB/OL] (2021- 11-24) [2023-12-30]. https://arxiv.org/abs/2111.12419.

[13] HOU Q, ZHOU D, FENG J. Coordinate attention for efficient mobile network design[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021: 13713-13722.

[14] WANG Q, WU B, ZHU P, et al. ECA-Net: Efficient channel attention for deep convolutional neural networks [C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020: 11534- 11542.

[1]	刘美岐, 金楷然, 李雅澜, 郭戈. 城市道路智能网联车辆轨迹鲁棒控制方法[J]. 交通运输系统工程与信息, 2024, 24(4): 31-40.
[2]	温惠英, 张昕怡, 黄俊达, 许鹏鹏. 考虑动态交互作用的智能车辆轨迹预测[J]. 交通运输系统工程与信息, 2024, 24(4): 60-68.
[3]	金盛, 周梦涛, 白聪聪. 多源数据驱动的隧道场景驾驶员识别方法[J]. 交通运输系统工程与信息, 2024, 24(4): 81-93.
[4]	赵朔, 奇格奇, 李培豪, 关伟. 基于脑电通道注意力机制的驾驶行为识别研究[J]. 交通运输系统工程与信息, 2024, 24(4): 283-291.
[5]	马千里, 王通, 邵帅, 贾鹏. 路测环境下无人驾驶小巴复杂场景运行评价[J]. 交通运输系统工程与信息, 2024, 24(4): 300-310.
[6]	董红召, 杨嘉炜, 全程. 车联网下公交专用道复用的动态清空控制方法[J]. 交通运输系统工程与信息, 2024, 24(3): 12-20.
[7]	刘江, 张楚, 蔡伯根, 王剑, 陆德彪. 基于样本增强的列车卫星定位伪距欺骗检测方法[J]. 交通运输系统工程与信息, 2024, 24(3): 32-42.
[8]	李建市, 徐友春, 齐尧, 谢德胜, 李华. 起伏道路条件下无人车纵向速度控制方法研究[J]. 交通运输系统工程与信息, 2024, 24(3): 43-52.
[9]	张金雷, 杨健, 刘晓冰, 陈瑶, 杨立兴, 高自友. 基于计算机视觉的轨道交通站内火灾检测与定位[J]. 交通运输系统工程与信息, 2024, 24(3): 53-63.
[10]	陈喜群, 朱奕璋, 谢宁珂, 耿茂思, 吕朝锋. 基于异构多智能体自注意力网络的路网信号协调顺序优化方法[J]. 交通运输系统工程与信息, 2024, 24(3): 114-126.
[11]	辛琪, 王嘉琪, 杨文科, 徐猛, 袁伟. 信控路段混行交通生态驾驶深度强化学习模型[J]. 交通运输系统工程与信息, 2024, 24(3): 127-139.
[12]	鲁文博, 张永, 李培坤, 王亭, 丛雅蓉. 自适应多视图融合图神经网络地铁客流预测模型[J]. 交通运输系统工程与信息, 2024, 24(3): 194-203.
[13]	倪浩原, 余贵珍, 李涵, 陈鹏, 刘喜, 王文达. 露天矿作业区无人矿车协同通行决策方法研究[J]. 交通运输系统工程与信息, 2024, 24(3): 277-289.
[14]	张磊, 彭金栓, 陈晓利. 接管场景驱动下非驾驶任务对驾驶人视觉与生理特征的影响[J]. 交通运输系统工程与信息, 2024, 24(3): 290-298.
[15]	魏丽英, 吴润泽. 基于鱼群效应的智能网联车队形成与演化机理研究[J]. 交通运输系统工程与信息, 2024, 24(2): 76-85.

无人机高空航拍视角下小尺度车辆精确检测方法

Accurate Detection Method of Small-scale Vehicles from Perspective of Unmanned Aerial Vehicle High-altitude Aerial Photography

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics