Unsafe driving behavior detection under complex lighting conditions

doi:10.11772/j.issn.1001-9081.2025020170

Journal of Computer Applications ›› 2026, Vol. 46 ›› Issue (2): 613-619.DOI: 10.11772/j.issn.1001-9081.2025020170

• Multimedia computing and computer simulation • Previous Articles

Unsafe driving behavior detection under complex lighting conditions

Quanjie LIU¹, Zhaoyi GU², Chunyuan WANG¹()

^1.School of Mechanical and Automotive Engineering，Qingdao University of Technology，Qingdao Shandong 266520，China
^2.School of Safety Engineering，Shenyang Aerospace University，Shenyang Liaoning 110136，China

Received:2025-02-24 Revised:2025-04-29 Accepted:2025-04-30 Online:2025-05-16 Published:2026-02-10
Contact: Chunyuan WANG
About author:LIU Quanjie， born in 2000， M. S. candidate. His research interests include object detection and tracking algorithms， deep learning.
GU Zhaoyi， born in 1999. His research interests include artificial intelligence， intelligent security.
WANG Chunyuan， born in 1980， M. S.， associate professor. His research interests include intelligent security， object detection algorithms. Email：chunyuanmail@163.com
Supported by:
National Natural Science Foundation of China(52474238)

复杂光照条件下的不安全驾驶行为检测

刘权捷¹, 顾兆一², 王春源¹()

^1.青岛理工大学机械与汽车工程学院，山东青岛 266520
^2.沈阳航空航天大学安全工程学院，沈阳 110136

通讯作者: 王春源
作者简介:刘权捷（2000—），男，辽宁抚顺人，硕士研究生，CCF会员，主要研究方向：目标检测跟踪算法、深度学习
顾兆一（1999—），男，辽宁大连人，主要研究方向：人工智能、智慧安防
王春源（1980—），男，山东诸城人，副教授，硕士，主要研究方向：智能安全、目标检测算法。Email：chunyuanmail@163.com
基金资助:
国家自然科学基金资助项目(52474238)

Abstract

Abstract:

For real-time and effective detection of unsafe driving behaviors such as not wearing seatbelts and using mobile phones under various complex lighting conditions， a deep learning-based unsafe driving behavior detection method under complex lighting conditions was designed. In the method， with YOLOv8n model selected as the foundation， a series of targeted improvements were implemented to enhance the detection performance. Firstly， a P6 scale was added to enable the model to capture the diversity of unsafe driving behaviors under various lighting conditions more comprehensively. Secondly， Spatial Separable Adaptive Convolution （SSAC） module was used to replace the traditional convolution module in the backbone network， thereby achieving lightweight design while improving feature extraction accuracy. Thirdly， Channel Prior Convolutional Attention （CPCA） was introduced to enhance the network’s focus on important features effectively and improve feature expression capability. Finally， the Selective Attention Feature Fusion （SAFF） structure was used to replace the original YOLOv8n neck network， thereby further improving the mode’s comprehensive performance. Experimental results show that compared to the original model， the improved YOLOv8n model increases the overall mean Average Precision （mAP） by 2.17%； under normal lighting conditions， the improvement is 1.76%； in night scenes， the improvement is 1.75%； in backlit environments， the improvement is 2.42%. Meanwhile， the improved YOLOv8n reaches 118 Frames Per Second （FPS） in comparison with other models （such as YOLO11n， RT-DETR（Real-Time DEtection TRansformer））， balancing precision and speed， demonstrating distinct advantages.

Key words: improved YOLOv8, unsafe driving behavior detection, deep learning, object detection algorithm, complex lighting condition

摘要：

为了在各种复杂光照条件下实时有效检测驾驶人员不系安全带和看手机等不安全行为，设计一种基于深度学习的复杂光照下不安全驾驶行为检测方法。该方法以YOLOv8n模型作为基础，实施一系列针对性的改进措施，以提升检测性能。首先，增加P6尺度，使模型能更全面地捕捉各种光照下不安全驾驶行为的多样性；其次，使用空间可分离自适应卷积（SSAC）模块替换主干网络的传统卷积模块，从而在提高特征提取精度的同时实现轻量化；再次，引入通道先验卷积注意力（CPCA），有效增强网络对重要特征的关注，并提升特征的表达能力；最后，使用选择注意特征融合（SAFF）结构替换原有YOLOv8n颈部网络，进一步提升模型的综合性能。实验结果表明，相较于原模型，改进后的YOLOv8n模型整体的平均精度均值（mAP）提升了2.17%；在正常光照条件下提升了1.76%；在夜间场景下提升了1.75%；在逆光环境下提升了2.42%。同时，改进后的YOLOv8n在与其他模型（如YOLO11n和RT-DETR（Real-Time DEtection TRansformer））的对比中，每秒帧数（FPS）达到118，精度与速度兼顾，展现出较明显的优势。

关键词: 改进YOLOv8, 不安全驾驶行为检测, 深度学习, 目标检测算法, 复杂光照条件

CLC Number:

TP391.9

Quanjie LIU, Zhaoyi GU, Chunyuan WANG. Unsafe driving behavior detection under complex lighting conditions[J]. Journal of Computer Applications, 2026, 46(2): 613-619.

刘权捷, 顾兆一, 王春源. 复杂光照条件下的不安全驾驶行为检测[J]. 《计算机应用》唯一官方网站, 2026, 46(2): 613-619.

Figures/Tables 10

References 19

[1]	MOHD AZAMI M F A， MISRO M Y， HAMIDUN R. Road crash dynamics in Malaysia： analysis of trends and patterns［J］. Heliyon， 2024， 10（18）： No.e37457.
[2]	KHOSRAVI E， HEMMATYAR A M A， SIAVOSHANI M J， et al. Safe Deep Driving behavior Detection （S3D）［J］. IEEE Access， 2022， 10： 113827-113838.
[3]	宁华晶. 不安全驾驶行为与道路交通事故之间的关系研究［D］. 兰州：兰州交通大学， 2023： 62-63.
	NING H J. Study on the relationship between unsafe driving behaviors and traffic crashes［D］. Lanzhou： Lanzhou Jiaotong University， 2023： 62-63.
[4]	LATTANZI E， FRESCHI V. Machine learning techniques to identify unsafe driving behavior by means of in-vehicle sensor data［J］. Expert Systems with Applications， 2021， 176： No.114818.
[5]	QU F， DANG N， FURHT B， et al. Comprehensive study of driver behavior monitoring systems using computer vision and machine learning techniques［J］. Journal of Big Data， 2024， 11： No.32.
[6]	KANG H B. Various approaches for driver and driving behavior monitoring： a review［C］// Proceedings of the 2013 IEEE International Conference on Computer Vision Workshops. Piscataway： IEEE， 2013： 616-623.
[7]	ABOSAQ H A， RAMZAN M， ALTHOBIANI F， et al. Unusual driver behavior detection in videos using deep learning models［J］. Sensors， 2022， 23（1）： No.311.
[8]	HU J， ZHANG X， MAYBANK S. Abnormal driving detection with normalized driving behavior data： a deep learning approach［J］. IEEE Transactions on Vehicular Technology， 2020， 69（7）： 6943-6951.
[9]	田文洪，曾柯铭，莫中勤，等. 基于卷积神经网络的驾驶员不安全行为识别［J］. 电子科技大学学报， 2019， 48（3）： 381-387.
	TIAN W H， ZENG K M， MO Z Q， et al. Recognition of unsafe driving behaviors based on convolutional neural network［J］. Journal of University of Electronic Science and Technology of China， 2019， 48（3）： 381-387.
[10]	骆文婕. 基于机器视觉的非安全驾驶行为检测算法研究［D］. 长沙：湖南大学， 2018： 26-27.
	LUO W J. Research on the detection algorithm of non-safe driving behavior based on machine vision［D］. Changsha： Hunan University， 2018： 26-27.
[11]	宋知霖. 基于深度学习的不安全驾驶行为检测算法研究［D］. 沈阳：沈阳工业大学， 2021： 22-27.
	SONG Z L. Research on unsafe driving behavior detection algorithm based on deep learning［D］. Shenyang： Shenyang University of Technology， 2021： 22-27.
[12]	REDMON J， DIVVALA S， GIRSHICK R， et al. You only look once： unified， real-time object detection［C］// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2016： 779-788.
[13]	李明煜，林家泉，高梅. 面向驾驶员面部目标的FD-YOLOv8轻量化检测算法［J］. 小型微型计算机系统， 2025， 46（11）： 2700-2707.
	LI M Y， LIN J Q， GAO M. FD-YOLOv8 lightweight detection algorithm for driver’s facial target［J］. Journal of Chinese Computer Systems， 2025， 46（11）： 2700-2707.
[14]	孙铁强，魏光辉，宋超，等. 基于YOLO的多模态钢轨表面缺陷检测方法［J］. 电子测量技术， 2024， 47（21）： 72-81.
	SUN T Q， WEI G H， SONG C， et al. Multi-modal rail surface defect detection method based on YOLO［J］. Electronic Measurement Technology， 2024， 47（21）： 72-81.
[15]	邱琳琳，朱卫纲，李永刚，等. 基于改进YOLOv8的SAR图像飞机目标检测算法［J］. 电光与控制， 2025， 32（3）： 101-110.
	QIU L L， ZHU W G， LI Y G， et al. Aircraft detection in SAR images based on improved YOLOv8［J］. Electronics Optics and Control， 2025， 32（3）： 101-110.
[16]	HUANG H， CHEN Z， ZOU Y， et al. Channel prior convolutional attention for medical image segmentation［J］. Computers in Biology and Medicine， 2024， 178： No.108784.
[17]	赵佰亭，程瑞丰，贾晓芬. 融合多尺度特征的YOLOv8裂缝缺陷检测算法［J］. 计算机工程与应用， 2024， 60（22）： 261-270.
	ZHAO B T， CHENG R F， JIA X F. YOLOv8 crack defect detection algorithm based on multi-scale features［J］. Computer Engineering and Applications， 2024， 60（22）： 261-270.
[18]	BOUHSISSIN S， SAEL N， BENABBOU F， et al. SafeSmartDrive： real-time traffic environment detection and driver behavior monitoring with machine and deep learning［J］. IEEE Access， 2024， 12： 169499-169517.
[19]	CHENG Q， LI H， YANG Y， et al. Towards efficient risky driving detection： a benchmark and a semi-supervised model［J］. Sensors， 2024， 24（5）： No.1386.

YOLOv8	P6融合SAFF	CPCA	SSAC	mAP	mAP_50：95	mAP₇₅	R	AP
YOLOv8	P6融合SAFF	CPCA	SSAC	mAP	mAP_50：95	mAP₇₅	R	抽烟	安全带	手机	食物
√				0.969	0.735	0.869	0.780	0.910	0.978	0.945	0.883
√	√			0.972	0.732	0.855	0.782	0.911	0.970	0.965	0.880
√		√		0.966	0.725	0.862	0.775	0.903	0.978	0.944	0.876
√			√	0.972	0.727	0.874	0.780	0.905	0.975	0.952	0.875
√	√	√		0.974	0.736	0.873	0.781	0.930	0.975	0.961	0.879
√	√		√	0.965	0.720	0.851	0.769	0.968	0.966	0.964	0.884
√		√	√	0.970	0.723	0.846	0.770	0.987	0.962	0.972	0.871
√	√	√	√	0.990	0.751	0.886	0.792	0.968	0.979	0.969	0.917

YOLOv8	P6融合SAFF	CPCA	SSAC	mAP	mAP_50：95	mAP₇₅	R	AP
YOLOv8	P6融合SAFF	CPCA	SSAC	mAP	mAP_50：95	mAP₇₅	R	抽烟	安全带	手机	食物
√				0.969	0.735	0.869	0.780	0.910	0.978	0.945	0.883
√	√			0.972	0.732	0.855	0.782	0.911	0.970	0.965	0.880
√		√		0.966	0.725	0.862	0.775	0.903	0.978	0.944	0.876
√			√	0.972	0.727	0.874	0.780	0.905	0.975	0.952	0.875
√	√	√		0.974	0.736	0.873	0.781	0.930	0.975	0.961	0.879
√	√		√	0.965	0.720	0.851	0.769	0.968	0.966	0.964	0.884
√		√	√	0.970	0.723	0.846	0.770	0.987	0.962	0.972	0.871
√	√	√	√	0.990	0.751	0.886	0.792	0.968	0.979	0.969	0.917

光照条件	模型	mAP	mAP_50：95	mAP₇₅	R
正常光照	YOLOv8n	0.968	0.754	0.880	0.791
正常光照	改进YOLOv8n	0.985	0.768	0.908	0.804
夜间条件	YOLOv8n	0.970	0.778	0.902	0.811
夜间条件	改进YOLOv8n	0.987	0.799	0.925	0.825
逆光条件	YOLOv8n	0.949	0.682	0.800	0.739
逆光条件	改进YOLOv8n	0.972	0.695	0.811	0.748

光照条件	模型	mAP	mAP_50：95	mAP₇₅	R
正常光照	YOLOv8n	0.968	0.754	0.880	0.791
正常光照	改进YOLOv8n	0.985	0.768	0.908	0.804
夜间条件	YOLOv8n	0.970	0.778	0.902	0.811
夜间条件	改进YOLOv8n	0.987	0.799	0.925	0.825
逆光条件	YOLOv8n	0.949	0.682	0.800	0.739
逆光条件	改进YOLOv8n	0.972	0.695	0.811	0.748

模型	mAP	mAP_50：95	mAP₇₅	R	GFLOPs	Params/10⁶	FPS
YOLOv3	0.973	0.702	0.829	0.756	282.0	103.67	37
YOLOv5n	0.972	0.735	0.866	0.773	7.1	2.50	134
YOLOv6n	0.971	0.729	0.867	0.774	7.9	2.67	98
YOLOv8n	0.969	0.735	0.869	0.780	8.1	3.00	129
YOLOv9t^［18］	0.970	0.738	0.867	0.777	8.5	3.17	130
YOLOv10n	0.963	0.721	0.872	0.776	8.2	3.08	131
YOLO11n	0.964	0.727	0.868	0.773	6.3	2.19	141
Gold-YOLO	0.974	0.732	0.859	0.768	10.5	6.02	107
YOLOv8-SPDConv	0.967	0.733	0.856	0.771	9.7	4.74	111
YOLOv8-EfficientViT	0.968	0.725	0.845	0.777	9.4	4.18	113
YOLOv8-StarNet	0.970	0.719	0.853	0.770	6.5	2.21	138
SSD-VGG16	0.962	0.731	0.859	0.764	98.0	28.17	64
Faster-RCNN-R50-FPN	0.974	0.732	0.812	0.699	208.0	34.39	45
Cascade-RCNN-R50-FPN	0.965	0.712	0.803	0.714	236.0	69.29	39
RetinaNet-R50-FPN	0.968	0.726	0.822	0.689	210.0	36.52	44
RT-DETR-R18	0.975	0.733	0.852	0.720	57.0	19.87	59
文献［8］模型	0.959	0.722	0.823	0.731	—	—	—
文献［10］模型	0.934	0.697	0.813	0.687	—	—	—
文献［11］模型	0.970	0.738	0.854	0.761	89.0	22.89	72
文献［19］模型	0.972	0.729	0.841	0.737	234.5	74.82	—
改进的YOLOv8n	0.990	0.751	0.886	0.792	9.1	4.06	118

Unsafe driving behavior detection under complex lighting conditions

复杂光照条件下的不安全驾驶行为检测

RichHTML

PDF

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 10

References 19

Related Articles 15

Recommended Articles

Metrics

[1]	Ming LI, Mengqi WANG, Aili ZHANG, Hua REN, Yuqiang DOU. Image steganography method based on conditional generative adversarial networks and hybrid attention mechanism [J]. Journal of Computer Applications, 2026, 46(2): 475-484.
[2]	Zeyi GUO, Fenglian LI, Lichun XU. Double decision mechanism-based deep symbolic regression algorithm [J]. Journal of Computer Applications, 2026, 46(2): 406-415.
[3]	Xiaolei CHEN, Zhiwei ZHENG, Xue HUANG, Zhenbin QU. Panoramic video super-resolution network combining spherical alignment and adaptive geometric correction [J]. Journal of Computer Applications, 2026, 46(2): 528-535.
[4]	Jinjiao LIN, Canshun ZHANG, Shuya CHEN, Tianxin WANG, Jian LIAN, Yonghui XU. Vehicle insurance fraud detection method based on improved graph attention network [J]. Journal of Computer Applications, 2026, 46(2): 437-444.
[5]	Haoqian JIANG, Dong ZHANG, Guanyu LI, Heng CHEN. SetaCRS： Conversational recommender system with structure-enhanced hierarchical task-oriented prompting strategy [J]. Journal of Computer Applications, 2026, 46(2): 368-377.
[6]	Feng HAN, Yongfeng BU, Haoxiang LIANG, Shuwen HUANG, Zhaoyang ZHANG, Shijie SUN. Vehicle trajectory anomaly detection based on multi-level spatio-temporal interaction dependency [J]. Journal of Computer Applications, 2026, 46(2): 604-612.
[7]	Xiaoyong BIAN, Peiyang YUAN, Qiren HU. Dual-coding space-frequency mixing method for infrared small target detection [J]. Journal of Computer Applications, 2026, 46(1): 252-259.
[8]	Zhihui ZAN, Yajing WANG, Ke LI, Zhixiang YANG, Guangyu YANG. Multi-feature fusion speech emotion recognition method based on SAA-CNN-BiLSTM network [J]. Journal of Computer Applications, 2026, 46(1): 69-76.
[9]	Hongjun ZHANG, Gaojun PAN, Hao YE, Yubin LU, Yiheng MIAO. Multi-source heterogeneous data analysis method combining deep learning and tensor decomposition [J]. Journal of Computer Applications, 2025, 45(9): 2838-2847.
[10]	Jin LI, Liqun LIU. SAR and visible image fusion based on residual Swin Transformer [J]. Journal of Computer Applications, 2025, 45(9): 2949-2956.
[11]	Bing YIN, Zhenhua LING, Yin LIN, Changfeng XI, Ying LIU. Emotion recognition method compatible with missing modal reasoning [J]. Journal of Computer Applications, 2025, 45(9): 2764-2772.
[12]	Weigang LI, Jiale SHAO, Zhiqiang TIAN. Point cloud classification and segmentation network based on dual attention mechanism and multi-scale fusion [J]. Journal of Computer Applications, 2025, 45(9): 3003-3010.
[13]	Zhixiong XU, Bo LI, Xiaoyong BIAN, Qiren HU. Adversarial sample embedded attention U-Net for 3D medical image segmentation [J]. Journal of Computer Applications, 2025, 45(9): 3011-3016.
[14]	Panfeng JING, Yudong LIANG, Chaowei LI, Junru GUO, Jinyu GUO. Semi-supervised image dehazing algorithm based on teacher-student learning [J]. Journal of Computer Applications, 2025, 45(9): 2975-2983.
[15]	Lina GE, Mingyu WANG, Lei TIAN. Review of research on efficiency of federated learning [J]. Journal of Computer Applications, 2025, 45(8): 2387-2398.