基于改进YOLOv8的矿卡司机疲劳驾驶检测

doi:10.16265/j.cnki.issn1003-3033.2025.01.0147

摘要/Abstract

摘要：

为解决露天矿区卡车司机疲劳驾驶检测漏检率和误检率高、鲁棒性较差的问题,构建基于改进YOLOv8的露天矿卡车司机疲劳驾驶检测模型(EBS-YOLO),提高疲劳检测的整体性能。首先,以YOLOv8为疲劳检测基础模型,通过添加小目标检测层,增强模型对小目标关注;其次,引入瓶颈注意力(BAM)模块,强化模型对小目标特征提取能力,尤其是对眼部特征提取能力;最后,将主干网络中跨阶段聚合模块(C2f)全部替换为高效多尺度注意力(EMA)模块,进而有效降低模型参数量和计算开销,以满足模型轻量化需求。结果表明: 改进后的YOLOv8模型检测效果较好,准确率、召回率、平均检测精度分别达到了93.6%、93.9%、96.5%,且模型内存大小仅有4.9 MB。相比于YOLOv8模型,改进后的模型能够快速准确识别出矿卡司机疲劳状态,满足实时性要求,从而有效预防疲劳驾驶事故发生。

关键词: 露天矿, 疲劳驾驶检测, 卡车司机, 小目标检测, YOLOv8

Abstract:

To address the high rates of missed detections and false alarms, as well as the poor robustness in fatigue driving detection for open-pit mine truck drivers, a fatigue driving detection model for mine truck drivers (EBS-YOLO) based on the improved YOLOv8 is constructed to enhance the overall performance of fatigue detection. Firstly, YOLOv8 was used as the basic model for fatigue detection, and a small target detection layer was added to enhance the model's attention to small targets. Secondly, the bottleneck attention module (BAM) was used to improve the model performance to extract small target features, especially eye features. Finally, all cross-stage aggregation modules (C2f) in the backbone network were replaced with efficient multi-scale attention (EMA) modules, thereby effectively reducing model parameters and computational overhead to meet the requirements of a lightweight model. The results showed that the improved YOLOv8 model had a great detection effect with the accuracy, recall rate, and average detection accuracy reaching 93.6%, 93.9%, and 96.5%, respectively, and the memory size of the model was only 4.9 MB. Compared with the YOLOv8 model, the improved model can quickly and accurately identify the fatigue state of mining truck drivers, meet real-time requirements, and effectively prevent fatigue-driving accidents.

Key words: open-pit mines, fatigue driving detection, truck driver, detection of small targets, YOLOv8

中图分类号:

X910安全人体学

顾清华, 殷书檀, 王丹, 李学现, 尹慧民. 基于改进YOLOv8的矿卡司机疲劳驾驶检测[J]. 中国安全科学学报, 2025, 35(1): 60-66.

GU Qinghua, YIN Shutan, WANG Dan, LI Xuexian, YIN Huimin. Mining truck driver fatigue driving detection based on improved YOLOv8[J]. China Safety Science Journal, 2025, 35(1): 60-66.

图/表 9

图1

表1

表2

表3

表4

图2

图3

表5

图4

参考文献 15

[1]	WANG Yi, HE Zhengxiang, WANG Liguang. Truck driver fatigue detection based on video sequences in open-pit mines[J]. Mathematics, Multidisciplinary Digital Publishing Institute, 2021, 9(22): 2908-2922.
[2]	胡剑锋, 王涛涛. 基于脑电信号模糊熵的驾驶疲劳检测分析[J]. 中国安全科学学报, 2018, 28(4): 13-18. doi: 10.16265/j.cnki.issn1003-3033.2018.04.003
	HU Jianfeng, WANG Taotao. Analysis of driving fatigue detection based on fuzzy entropy of EEG signals[J]. China Safety Science Journal, 2018, 28(4): 13-18. doi: 10.16265/j.cnki.issn1003-3033.2018.04.003
[3]	ZHANG Pengbo, WANG Xue, CHEN Junfeng, et al. Feature weight driven interactive mutual information modeling for heterogeneous bio-signal fusion to estimate mental workload[J]. Sensors, 2017, 17(10): 2315-2339.
[4]	WANG Fuwang, WANG Hong, FU Rongrong. Real-time ECG-based detection of fatigue driving using sample entropy[J]. Entropy, 2018, 20(3): 196-212.
[5]	程文冬, 付锐, 袁伟, 等. 驾驶人疲劳监测预警技术研究与应用综述[J]. 中国安全科学学报, 2013, 23(1): 155-160.
	CHENG Wendong, FU Rui, YUAN Wei, et al. Overview of researches on driver fatigue monitoring and prewarning technologies and their applications[J]. China Safety Science Journal, 2013, 23(1): 155-160.
[6]	ZHANG Jun, WU Zhongcheng, LI Fang, et al. A deep learning framework for driving behavior identification on in-vehicle can-bus sensor data[J]. Sensors, Multidisciplinary Digital Publishing Institute, 2019, 19(6): 1356-1373.
[7]	辛嵩, 宋明达, 王泽明, 等. 针对特定驾驶员的疲劳驾驶检测方法[J]. 安全与环境学报, 2023, 23(1): 147-152.
	XIN Song, SONG Mingda, WANG Zeming, et al. Fatigue driving detection method for specific drivers[J]. Journal of Safety and Environment, 2023, 23(1): 147-152.
[8]	MAGAN E, SESMERO M P, ALONSOWEBER J M, et al. Driver drowsiness detection by applying deep learning techniques to sequences of images[J]. Applied Sciences, 2022, 12(3): 1145-1170.
[9]	HUANG Rui, WANG Yan, LI Zijian, et al. RF-DCM: multi-granularity deep convolutional model based on feature recalibration and fusion for driver fatigue detection[J]. IEEE Transactions on Intelligent Transportation Systems, 2022, 23(1): 630-640.
[10]	杨艳艳, 李雷孝, 林浩. 提取驾驶员面部特征的疲劳驾驶检测研究综述[J]. 计算机科学与探索, 2023, 17(6): 1249-1267. doi: 10.3778/j.issn.1673-9418.2208041
	YANG Yanyan, LI Leixiao, LIN Hao. A review of fatigue driving detection research based on extracting driver's facial features[J]. Computer Science and Exploration, 2023, 17(6): 1249-1267.
[11]	LI Kening, GONG Yunbo, REN Zijiang. A fatigue driving detection algorithm based on facial multi-feature fusion[J]. IEEE Access, 2020, 8(8): 101 244-101 259.
[12]	胡习之, 黄冰瑜. 基于面部特征分析的疲劳驾驶检测方法[J]. 科学技术与工程, 2021, 21(4): 1629-1636.
	HU Xizhi, HUANG Bingyu. Fatigue driving detection system based on face feature analysis[J]. Science Technology and Engineering, 2021, 21 (4): 1629-1636.
[13]	LIU Mingzhou, XU Xin, HU Jing, et al. Real time detection of driver fatigue based on CNN-LSTM[J]. IET Image Processing, 2022, 16(2): 576-595. doi: 10.1049/ipr2.12373
[14]	方浩杰, 董红召, 林少轩, 等. 多特征融合的驾驶员疲劳状态检测方法[J]. 浙江大学学报:工学版, 2023, 57(7): 1287-1296.
	FANG Haojie, DONG Hongzhao, LIN Shaoxuan, et al. Driver fatigue state detection method based on multi-feature fusion[J]. Journal of Zhejiang University:Engineering Science, 2023, 57(7): 1287-1296.
[15]	刘子洋, 徐慧英, 朱信忠, 等. Bi-YOLO:一种基于YOLOv8改进的轻量化目标检测算法[J]. 计算机工程与科学, 2023, 46(8): 1444-1454.
	LIU Ziyang, XU Huiying, ZHU Xinzhong, et al. Bi-YOLO: an improved lightweight object detection algorithm based on YOLOV8[J]. Computer Engineering and Science, 2023, 46(8): 1444-1454.

模型	mAP @0.5/%	Params/ M	FLOPs/ G	模型大小/ MB
YOLOv8n	91.2	3.01	8.1	6.3
小目标	94.1	10.95	37.8	22.1
小目标+Biformer	94.8	10.98	38.4	22.4
小目标+GAM	94.3	11.28	38.1	22.5
小目标+SimAM	94.6	10.94	37.8	22.3
小目标+CBAM	94.5	11.62	39.2	23.1
小目标+BAM	96.3	11.53	38.9	22.3

序号	小目标	BAM	EMA	mAP@0.5/%	mAP@0.5-0.95/%	Params/M	FLOPs/G	模型大小/MB
1	—	—	—	91.2	53.2	3.01	8.1	6.3
2	√	—	—	94.9	54.9	10.95	37.8	22.1
3	√	√	—	96.3	56.1	11.53	38.9	22.3
4	√	√	√	96.5	56.7	2.28	10.7	4.9

模型	mAP @0.5/%	Params/ M	FLOPs/ G	FPS/ (帧·s^-1)	模型大小/MB
YOLOv8n	91.2	3.01	8.1	51	6.3
YOLOv3-tiny	85.3	105.7	292.2	5	206.8
YOLOv5s	91.4	4.9	14.9	44	13.2
YOLOv6n	89.5	4.5	12.8	43	8.9
YOLOv7-tiny	90.5	6.7	15.1	41	72.6
SSD	93.8	26.4	107.8	25	93.5
Fast-RCNN	92.3	42.6	134.5	10	106.5
EBS-YOLO	96.5	2.28	10.7	49	4.9

模型	人脸	睁眼	闭眼	张嘴	闭嘴	打哈欠	戴口罩	低头	P/%	R/%	mAP@0.5/%
YOLOv8n	97.9	85.6	85.8	89.3	87.7	92.2	94.6	96.5	88.4	87.8	91.2
EBS-YOLO	98.5	95.4	96.2	96.6	95.9	96.5	97.1	95.8	93.6	93.9	96.5

模型	P/ %	R/ %	mAP @0.5/%	FPS/ (帧·s^-1)	模型大小/MB
YOLOv8n	91.4	91.0	92.4	52	6.3
YOLOv5s	92.1	85.0	91.9	46	13.2
YOLOv6n	84.1	88.2	89.2	44	8.9
YOLOv7-tiny	90.7	90.4	94.2	40	72.2
SSD	94.0	83.6	94.1	28	93.5
Fast-RCNN	92.9	82.1	93.3	19	106.5
EBS-YOLO	96.8	96.5	97.4	49	4.9