基于深度学习模型的疲劳驾驶行为识别算法

doi:10.3969/j.issn.1674-8484.2024.01.013

汽车安全与节能学报 ›› 2024, Vol. 15 ›› Issue (1): 121-128.DOI: 10.3969/j.issn.1674-8484.2024.01.013

基于深度学习模型的疲劳驾驶行为识别算法

张海民()

安徽信息工程学院计算机与软件工程学院，芜湖 241000，中国

收稿日期:2023-05-06 修回日期:2023-09-14 出版日期:2024-02-29 发布日期:2024-02-29
作者简介:张海民（1983—），男（汉），安徽，副教授。E-mail：2019000343@aiit.edu.cn
基金资助:
安徽省高校自然科学重点研究项目(KJ2021A1206);安徽高等学校省级质量工程项目(2022zygzts053);安徽省哲学社会科学规划项目(AHSKY2021D142)

Algorithm of fatigue driving behavior recognition based on deep learning model

ZHANG Haimin()

Anhui Institute of Information Technology, School of Computer and Software Engineering, Wuhu 241000, China

Received:2023-05-06 Revised:2023-09-14 Online:2024-02-29 Published:2024-02-29

摘要/Abstract

摘要：

为降低道路交通事故发生率，提出了一种基于深度学习模型的疲劳驾驶行为识别算法。采用照度增强和反射分量均衡化的方法，以提高视频图像质量。将机器视觉工具箱软件用于提取疲劳驾驶人脸行为特征，并通过双流网络构建和训练深度学习模型，实现对疲劳驾驶行为识别。选择了不同睡眠时间段参与者在全封闭路段内的驾驶行为图像，作为实验测试目标。结果表明：用该算法测试1 000张疲劳驾驶行为图像时，识别时间为89 ms，精准度为97.6 %，召回率为97.0 %；算力需求（每秒所执行的浮点运算次数，FLOPS） ≤ 88；该算法能够提高疲劳驾驶行为的识别精度，有助于降低道路交通事故的发生率。

关键词: 疲劳驾驶, 行为识别, 深度学习模型, 图像增强, 特征提取

Abstract:

A fatigue driving behavior recognition algorithm was proposed based on the deep learning model to identify fatigue driving behavior and to reduce the incidences of road traffic accidents. An illuminance enhancement method and a reflection component equalization method were used to improve the quality of video images. The Machine Vision Toolbox software was used to extract facial behavior features of fatigued drivers. A deep learning model was constructed and trained by using a dual-stream network to achieve fatigue driving behavior recognition. The images of participants' driving behavior in fully enclosed segments during different sleep periods were selected as experimental test targets. The results show that the recognition time is 89 ms, the accuracy is 97.6%, and the recall rate is 97.0% when 1 000 images of fatigue driving behavior are tested by the proposed algorithm; The computing power requirement (floating-point operations per second, FLOPS) is less than or equal to 88. Therefore, this algorithm improves the recognition accuracy of fatigue driving behavior, helps to reduce the incidence of road traffic accidents.

Key words: fatigue driving, behavior recognition, deep learning model, image enhancement, feature extraction

中图分类号:

TP274

张海民. 基于深度学习模型的疲劳驾驶行为识别算法[J]. 汽车安全与节能学报, 2024, 15(1): 121-128.

ZHANG Haimin. Algorithm of fatigue driving behavior recognition based on deep learning model[J]. Journal of Automotive Safety and Energy, 2024, 15(1): 121-128.

图/表 7

参考文献 15

[1]	任俊, 魏霞, 黄德启, 等. 基于眼睛状态多特征融合的疲劳驾驶检测[J]. 计算机工程与设计, 2022, 43(11): 3187-3194.
	REN Jun, WEI Xia, HUANG Deqi, et al. Fatigue driving detection based on eye state multi-feature fusion[J]. Compu Eng Des, 2022, 43(11): 3187-3194. (in Chinese)
[2]	徐少平, 张贵珍, 林珍玉, 等. 一种多图像局部结构化融合的低照度图像增强算法[J]. 自动化学报, 2022, 48(12): 2981-2995.
	XU Shaoping, ZHANG Guizhen, LIN Zhenyu, et al. A low-illumination image enhancement algorithm based on multi-image local structured fusion[J]. Acta Automatica, 2022, 48(12): 2981-2995. (in Chinese)
[3]	Verma A, Meenpal T, Acharya B. Human body pose distance image analysis for action recognition[J]. Int’l J Pattern Recog Artif Intel, 2022, 36(7): 2255012. 1-2255012.20.
[4]	胡峰松, 程哲坤, 徐青云, 等. 基于多特征融合的疲劳驾驶状态识别方法研究[J]. 湖南大学学报(自然科学版), 2022, 49(4): 100-107.
	HU Fengsong, CHENG Zhekun, XU Qingyun, et al. Research on Fatigue Driving State Recognition Based on Multi-feature Fusion[J]. J Hunan Univ (Nat Sci), 2022, 49(4): 100-107. (in Chinese)
[5]	DU Zhouning, Mukaidani H. Linear dynamical systems approach for human action recognition with dual-stream deep features[J]. Appl Intell, 2021, 52(1): 452-470. doi: 10.1007/s10489-021-02367-6
[6]	田垚, 李建良, 郭秋蕊, 等. 多特征因素的疲劳驾驶检测方法[J]. 天津科技大学学报, 2022, 37(2): 29-34.
	TIAN Yao, LI Jianliang, GUO Qiurei, et al. Fatigue driving detection method based on multi-characteristic factors[J]. J Tianjin Univ Sci Tech, 2022, 37(2): 29-34. (in Chinese)
[7]	巩霞, 魏浩然. 基于图像增强的纯蓝墨水手写档案修复仿真[J]. 计算机仿真, 2022, 39(10): 481-485.
	GONG Xia, WEI Haoran. Simulation of handwritten file restoration with pure blue ink based on image enhancement[J]. Computer Simu, 2022, 39(10): 481-485. (in Chinese)
[8]	林俊彦, 刘春晓, 章金凯, 等. 面向纹理平滑的方向性滤波尺度预测模型[J]. 中国图象图形学报, 2022, 27(8): 2506-2515.
	LIN Junyan, LIU Chunxiao, ZHANG Jinkai, et al. Directional filter scale prediction model for texture smoothing[J]. Chin J Image Graph, 2022, 27(8): 2506-2515. (in Chinese)
[9]	毕秀丽, 邱雨檬, 肖斌, 等. 基于统计特征的图像直方图均衡化检测方法[J]. 计算机学报, 2021, 44(2): 292-303.
	BI Xiuli, QIU Yumeng, XIAO Bin, et al. Image Histogram Equalization Detection Based on Statistical Features[J]. J Compu Sci, 2021, 44(2): 292-303. (in Chinese)
[10]	XU Canran, PENG Zizhao, HU Xuanzhen, et al. FPGA-Based low-visibility enhancement accelerator for video sequence by adaptive histogram equalization with dynamic clip-threshold[J]. IEEE Trans Circuits Syst, I: Regular Papers, 2020, 67(11): 3954-3964.
[11]	李小平, 白超. 基于深度学习的司机疲劳驾驶检测方法研究[J]. 铁道学报, 2021, 43(6): 78-87.
	LI Xiaoping, BAI Chao. Research on driver fatigue detection based on deep learning[J]. J Railways, 2021, 43(6): 78-87. (in Chinese)
[12]	张海民, 程菲. 深度学习下盲人避撞路径导航方法研究[J]. 南京信息工程大学学报, 2022, 14(2): 220-226.
	ZHANG Haimin, CHENG Fei. Research on Blind Collision Avoidance Navigation Based on Deep Learning[J]. J Nanjing Univ Info Sci Tech, 2022, 14(2): 220-226. (in Chinese)
[13]	张闯, 朱天军, 李学民. 基于深度学习和面部多特征融合的驾驶员疲劳检测研究[J]. 计算机测量与控制, 2022, 30(12): 42-50.
	ZHANG Chuang, ZHU Tianjun, LI Xuemin. Driver fatigue detection based on deep learning and multi-feature fusion[J]. Compu Meas Control, 2022, 30(12): 42-50. (in Chinese)
[14]	张福玲, 张少敏, 支力佳, 等. 融合注意力机制和特征金字塔网络的CT图像肺结节检测[J]. 中国图象图形学报, 2021, 26(9): 2156-2170.
	ZHANG Fuling, ZHANG Shaomin, ZHI Lijia, et al. Detection of pulmonary nodules in CT images by fusion of attention mechanism and characteristic pyramid network[J]. Chin J Image Graph, 2021, 26(9): 2156-2170. (in Chinese)
[15]	陈广秋, 陈昱存, 李佳悦, 等. 基于DNST和卷积稀疏表示的红外与可见光图像融合[J]. 吉林大学学报(工学版), 2021, 51(3): 996-1010.
	CHEN Guangqiu, CHEN Yucun, LI Jiayue, et al. Infrared and visible image fusion based on DNST and convolution sparse representation[J]. J Jilin Univ (Eng Tech Ed), 2021, 51(3): 996-1010. (in Chinese)

	详细情况
年龄范围	20~30岁
性别	5名男性、5名女性
睡眠时间段	23:00~06:00、02:00~08:00、22:00~06:00、9:00~05:00

	详细情况
年龄范围	20~30岁
性别	5名男性、5名女性
睡眠时间段	23:00~06:00、02:00~08:00、22:00~06:00、9:00~05:00

测试图像数量	方法编号	识别时间 / ms	精准度 / %	召回率 / %
200	1	15.00	99.00	100.00
	2	18.00	98.66	99.85
	3	20.00	98.37	98.98
	4	19.00	98.52	99.42
	5	18.50	98.44	98.20
400	1	30.00	99.00	99.88
	2	35.00	98.91	99.02
	3	42.00	98.06	98.99
	4	38.50	98.49	99.01
	5	39.25	98.27	98.00
600	1	45.00	99.68	98.86
	2	53.00	98.79	98.16
	3	61.00	98.24	97.55
	4	57.00	98.52	97.86
	5	58.00	97.38	96.70
800	1	62.00	98.73	97.96
	2	70.00	97.45	97.18
	3	84.00	96.83	96.54
	4	77.00	97.14	96.86
	5	79.50	95.99	95.70
1 000	1	89.00	97.64	97.04
	2	98.00	96.45	95.57
	3	117.00	95.39	93.78
	4	107.50	95.92	94.68
	5	111.25	94.66	93.23

测试图像数量	方法编号	识别时间 / ms	精准度 / %	召回率 / %
200	1	15.00	99.00	100.00
	2	18.00	98.66	99.85
	3	20.00	98.37	98.98
	4	19.00	98.52	99.42
	5	18.50	98.44	98.20
400	1	30.00	99.00	99.88
	2	35.00	98.91	99.02
	3	42.00	98.06	98.99
	4	38.50	98.49	99.01
	5	39.25	98.27	98.00
600	1	45.00	99.68	98.86
	2	53.00	98.79	98.16
	3	61.00	98.24	97.55
	4	57.00	98.52	97.86
	5	58.00	97.38	96.70
800	1	62.00	98.73	97.96
	2	70.00	97.45	97.18
	3	84.00	96.83	96.54
	4	77.00	97.14	96.86
	5	79.50	95.99	95.70
1 000	1	89.00	97.64	97.04
	2	98.00	96.45	95.57
	3	117.00	95.39	93.78
	4	107.50	95.92	94.68
	5	111.25	94.66	93.23

基于深度学习模型的疲劳驾驶行为识别算法

Algorithm of fatigue driving behavior recognition based on deep learning model

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 7

参考文献 15

相关文章 6

编辑推荐

Metrics

本文评价

期刊信息

在线期刊

作者中心

审稿中心

联系我们

[1]	刘延伟, 黄志明, 高博麟, 钟薇, 陈嘉星, 刘家熙. 车载视角下基于视觉信息的前车行为识别[J]. 汽车安全与节能学报, 2023, 14(6): 707-714.
[2]	宋巍, 张光德. 基于改进EfficientDet网络的疲劳驾驶状态检测方法[J]. 汽车安全与节能学报, 2022, 13(4): 651-658.
[3]	朱艳, 谢忠志, 于雯, 李曙生, 张逊. 低光照环境下基于面部特征点的疲劳驾驶检测技术[J]. 汽车安全与节能学报, 2022, 13(2): 282-289.
[4]	屈贤，余烽，赵悦. 基于熵权灰色关联和 D-S 证据理论的疲劳驾驶险态辨识[J]. JASE, 2018, 9(2): 164-170.
[5]	徐军莉，闵建亮，胡剑锋，王平. 定性推理生成器在驾驶疲劳检测中的应用[J]. JASE, 2018, 9(01): 32-40.
[6]	马添翼, 成波. 基于面部表情特征的驾驶员疲劳状态识别[J]. 汽车安全与节能学报, 2010, 1(3): 200-204.