基于YOLOv10-vehicle算法的复杂天气情况下车辆目标检测方法

doi:10.13229/j.cnki.jdxbgxb.20250016

摘要/Abstract

摘要：

针对在阴天、雨天、夜晚等复杂天气情况下，车辆目标检测受到光照、雨雪、沙尘等因素影响，出现错检、漏检等问题，提出了一种YOLOv10-vehicle目标检测算法。首先，设计了一种新的注意力机制模块WT-PSA，在复杂天气下提高模型对车辆目标的关注度；其次，改进了SPPF模块，通过引入平均池化操作，改善因最大池化操作导致的特征信息提取不足问题；然后，设计了改进的C2f-OD模块，提升主干网络提取图片特征信息的能力；最后，将模型损失函数替换为Focal EIoU，以加快收敛速度并降低损失值；在车辆数据集UA-DETRAC上进行对比实验，改进后的算法平均准确率（mAP@0.5）相较原算法提升了5.1%，证明了YOLOv10-vehicle算法在复杂天气、恶劣天气下车辆检测方面的优越性。同时，在VOC公共数据集上进行实验验证，YOLOv10-vehicle算法在检测车辆目标时检测精度提高了2.8%，证明了本文改进算法的泛化性。

关键词: 车辆目标检测, YOLOv10n, C2f模块, 恶劣天气

Abstract:

In the face of complex weather conditions such as cloudy days， rainy days， and nights， vehicle target detection is affected by factors such as lighting， rain， snow， and dust. As a result， problems like false detections and missed detections occur. To address these issues， a YOLOv10-vehicle target detection algorithm is proposed. Firstly， a new attention mechanism module named WT-PSA is designed to improve the model's attention to vehicle targets under complex weather. Secondly， the SPPF module is improved by introducing the average pooling operation to address the problem of insufficient feature information extraction caused by the max pooling operation. Then， an improved C2f-OD module is put forward to enhance the ability of the backbone network to extract image feature information. Finally， the model's loss function is replaced with Focal EIoU to accelerate the convergence speed and reduce the loss value. Comparative experiments are conducted on the vehicle dataset UA-DETRAC. The mean average precision （mAP@0.5） of the improved algorithm is increased by 5.1% compared with that of the original algorithm， demonstrating the superiority of the YOLOv10-vehicle algorithm in vehicle detection under complex and severe weather conditions. Meanwhile， experiments are also carried out on the VOC public dataset. The detection accuracy of the YOLOv10-vehicle algorithm in detecting vehicle targets is improved by 2.8%， which verifies the generalization ability of the improved algorithm in this paper.

Key words: vehicle target detection, YOLOv10n, C2f module, severe weather

中图分类号:

TP391.4

杜宏,顾宸瑜,张孝峥,刘高天,李兴鑫,杨忠琳. 基于YOLOv10-vehicle算法的复杂天气情况下车辆目标检测方法[J]. 吉林大学学报(工学版), 2025, 55(10): 3309-3318.

Hong DU,Chen-yu GU,Xiao-zheng ZHANG,Gao-tian LIU,Xing-xin LI,Zhong-lin YANG. Vehicle target detection method based on the YOLOv10-vehicle algorithm under complex weather conditions[J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(10): 3309-3318.

图/表 15

图1

图2

图3

图4

图5

图6

图7

图8

表1

表2

图9

表3

表4

图10

表5

参考文献 22

[1]	赵奇慧, 刘艳洋, 项炎平. 基于深度学习的单阶段车辆检测算法综述[J]. 计算机应用, 2020, 40(): 30-36.
	Zhao Qi-hui, Liu Yan-yang, Xiang Yan-ping. A review of single-stage vehicle detection algorithms based on deep learning[J]. Journal of Computer Applications, 2020, 40 (Sup.2): 30-36.
[2]	梁鸿, 王庆玮, 张千, 等.小目标检测技术研究综述[J]. 计算机工程与应用, 2021, 57(1): 17-28.
	Liang Hong, Wang Qing-wei, Zhang Qian, et al. A review of research on small object detection technology[J]. Computer Engineering and Applications, 2021, 57(1): 17-28.
[3]	Simonyan K, Zisserman A.Very deep convolutional networks for large-scale image recognition[J]. Computer Science, 2014, 21(8): 91-103.
[4]	Krizhevsky A, Sutskever I, Hinton E G. ImageNet classification with deep convolutional neural networks[J]. Communications of the ACM, 2017, 60(6): 84-90.
[5]	Ren S Q, He K M, 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.
[6]	薛珊, 王亚博, 吕琼莹, 等. 基于YOLOX-drone的反无人机系统抗遮挡目标检测算法[J]. 工程科学学报, 2023, 45(9): 1539-1549.
	Xue Shan, Wang Ya-bo, Qiong-ying Lü, et al. Anti-occlusion target detection algorithm for Anti-UAV systems based on YOLOX-drone[J]. Journal of University of Science and Technology Beijing, 2023, 45(9): 1539-1549.
[7]	薛珊, 卢涛, 吕琼莹, 等. 基于多尺度融合和轻量化网络的无人机目标检测算法[J]. 湖南大学学报: 自然科学版, 2023, 50(8): 82-93.
	Xue Shan, Lu Tao, Qiong-ying Lyu, et al. UAV Target detection algorithm based on multi-scale fusion and lightweight network [J]. Journal of Hunan University(Natural Sciences), 2023, 50(8): 82-93.
[8]	薛珊, 张亚亮, 吕琼莹, 等. 复杂背景下的反无人机系统目标检测算法[J]. 吉林大学学报: 工学版, 2023,53(3): 891-901.
	Xue Shan, Zhang Ya-liang, Qiong-ying LÜ, et al. Target detection algorithm for Anti-UAV systems in complex backgrounds[J]. Journal of Jilin University (Engineering and Technology Edition), 2023, 53(3): 891-901.
[9]	周飞, 郭杜杜, 王洋, 等. 基于改进YOLOv8的交通监控车辆检测算法[J]. 计算机工程与应用, 2024, 60(6): 110-120.
	Zhou Fei, Guo Du-du, Wang Yang, et al. Traffic monitoring vehicle detection algorithm based on improved YOLOv8[J]. Computer Engineering and Applications, 2024, 60(6): 110-120.
[10]	许德刚, 王双臣, 王再庆, 等. 改进YOLOv8算法的城市车辆目标检测 [J]. 计算机工程与应用, 2024, 60(18): 136-146.
	Xu De-gang, Wang Shuang-chen, Wang Zai-qing, et al. Urban Vehicle target detection algorithm based on improved YOLOv8[J]. Computer Engineering and Applications, 2024, 60(18): 136-146.
[11]	张利丰, 田莹. 改进YOLOv8的多尺度轻量型车辆目标检测算法[J]. 计算机工程与应用, 2024, 60(3): 129-137.
	Zhang Li-feng, Tian Ying. An improved YOLOv8-based multi-scale lightweight vehicle object detection algorithm[J]. Computer Engineering and Applications, 2024, 60(3): 129-137.
[12]	Redmon J, Divvala S, Girshick R, et al. You only look once: unified, real-time object detection[C]∥IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, USA, 2016: 779-788.
[13]	Redmon J, Farhadi A. YOLO9000: better, faster, stronger[C]∥IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, USA, 2017: 6517-6525.
[14]	Redmon J, Farhadi A. YOLOv3: an incremental improvement[C]∥IEEE Conference on Computer Vision and Pattern Recognition, Salt Lake City, USA, 2018:774-782.
[15]	Bochkovskiy A, Wang C Y, Liao H. YOLOv4: optimal speed and accuracy of object detection[J/OL].[2020-04-23]. .
[16]	薛珊, 陈宇超, 吕琼莹, 等. 基于双支路神经网络的无人机图像识别方法[J]. 计算机仿真, 2023, 40(7): 233-238.
	Xue Shan, Chen Yu-chao, Qiong-ying Lü, et al. UAV Image recognition method based on dual-branch neural network[J]. Computer Simulation, 2023, 40(7): 233-238.
[17]	Wang A, Chen H, Liu L H, et al. YOLOv10: real-time end-to-end object detection[J/OL].[2020-04-23]. .
[18]	Ben-Hamu H, Wolf L, Polyak A. Wavelet convolutions for large receptive fields[J/OL].[2020-04-23]. .
[19]	Yao A, Shen Y, Guo Y. Omni-dimensional dynamic convolution[J/OL].[2020-04-23]. .
[20]	Zhang Y, Ren W, Zhang Z, et al. Focal and efficient IOU loss for accurate bounding box regression[J/OL].[2020-04-23]. .
[21]	薛珊, 张振, 吕琼莹, 等. 基于卷积神经网络的反无人机系统图像识别方法[J]. 红外与激光工程, 2020, 49(7): 250-257.
	Xue Shan, Zhang Zhen, Qiong-ying Lü, et al. Image recognition method of anti-UAV system based on convolutional neural network[J]. Infrared and Laser Engineering, 2020, 49(7): 250-257.
[22]	Jocher G, Chaurasia A, Qiu J. YOLO by ultralytics[J/OL].[2020-04-23]..

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

环境项	环境规格
CPU	Intel（R） Core（TM） i5-13600KF
内存	32 GB
显卡	NVIDIA RTX 4070
操作系统	Windows 11
编程语言	Python 3.9.19
深度学习框架	Pytorch 2.4.0
集成开发环境	Pycharm 社区版
CUDA	12.1
CUDNN	9.0.0

超参数	系数值
初始学习率η₀	0.01
循环学习率η	0.01
动量β	0.937
批次大小	16
图片尺寸/像素	640×640
训练轮次	100
预热学习轮数	3.0
预热训练动量	0.8
IoU训练阈值	0.7

实验序号	WT-PSA	C2f-OD	SPPF_A	Focal EIoU	Params/10⁶	FLOPs	mAP 0.5%
1	×	×	×	×	2.70	8.3	82.7
2	√	×	×	×	3.33	8.8	83.9
3	√	√	×	×	3.70	9.0	86.5
4	√	√	√	×	3.75	9.1	86.9
5	√	√	√	√	3.75	9.1	87.8

Model	Params/10⁶	GFLOPS	mAP 0.5%	mAP0.5%~0.95%
Faster-RCNN	36.5	23.3	93.8	60.4
YOLOv7-tiny	6.01	13.2	82.9	54.7
YOLOv8n	3.01	8.2	78.7	50.0
YOLOv10n	2.70	8.3	82.7	57.6
YOLOv10s	8.04	24.6	86.7	59.2
YOLOv11n	2.58	6.3	78.3	50.2
Ours	3.75	9.1	87.8	60.2

Model	mAP@0.5%	mAP@0.5-Car%
YLOLv10n	37.7	52.4
YOLOv10-vehicle	37.3	55.2