Identification model of unsafe behaviors among operators in machining workshops

doi:10.16265/j.cnki.issn1003-3033.2025.08.0176

Abstract

Abstract:

To enhance the safety management of operators in machining workshops, an identification model based on YOLOv11 was constructed. The YOLOv11 model was improved by integrating the MetaFormer architecture, Mixed Aggregation Network (MANet) module, and Adaptive Feature Grid Convolution Attention (AFGC Attention) mechanism. A video dataset captured in a real workshop environment was established to validate the identification model. The results show that the improved YOLOv11 model can identify three types of behaviors, namely unattended operation, operating without a face shield, and operating without protective clothing, with F₁scores exceeding 0.93 for all categories. The improved model demonstrates a significant enhancement in identifying small-sized targets, with the F₁ score for identifying glove-wearing behavior increasing from 0.684 to 0.708, and the mAP@0.5 value rising from 0.604 to 0.651. The research findings may provide technical support for the identification and early warning of unsafe behaviors among operators in machining workshops.

Key words: machining workshops, operators, unsafe behaviors, YOLOv11, attention mechanism

CLC Number:

X941

HU Xiaofeng, TENG Teng, HU Jinming, WEN Jiajun. Identification model of unsafe behaviors among operators in machining workshops[J]. China Safety Science Journal, 2025, 35(8): 40-47.

Figures/Tables 8

Fig.1

Fig.2

Fig.3

Table 1

Table 2

Table 3

Table 4

Table 5

References 15

[1]	YANG Meng, WU Chengke, GUO Yuanjun, et al. Transformer-based deep learning model and video dataset for unsafe action identification in construction projects[J]. Automation in Construction, 2023, 146: DOI: 10.1016/j.autcon.2022.104703.
[2]	佟瑞鹏, 张艳伟. 人工智能技术在矿工不安全行为识别中的融合应用[J]. 中国安全科学学报, 2019, 29(1):7-12. doi: 10.16265/j.cnki.issn1003-3033.2019.01.002
	TONG Ruipeng, ZHANG Yanwei. Integration between artificial intelligence technologies for miners' unsafe behavior identification[J]. China Safety Science Journal, 2019, 29(1): 7-12. doi: 10.16265/j.cnki.issn1003-3033.2019.01.002
[3]	赵荣泳, 韦炳宇, 朱文杰, 等. 公共场所行人异常行为识别方法综述[J]. 中国安全科学学报, 2024, 34(2):83-93. doi: 10.16265/j.cnki.issn1003-3033.2024.02.1125
	ZHAO Rongyong, WEI Bingyu, ZHU Wenjie, et al. Overview of recognition methods of pedestrian abnormal behaviors in public places[J]. China Safety Science Journal, 2024, 34(2): 83-93. doi: 10.16265/j.cnki.issn1003-3033.2024.02.1125
[4]	REDMON J, DIVVALA S, GIRSSHICK R, et al. You only look once: unified, real-time object detection[C]. Proceeding of the IEEE Conference on Computer Vision and Pattern Recognition, IEEE, 2016:779-788.
[5]	LI Lijuan, ZHANG Peng, YANG Shipin, et al. YOLOv5-SFE: an algorithm fusing spatio-temporal features for detecting and recognizing workers' operating behaviors[J]. Advanced Engineering Informatics, 2023, 56: DOI: 10.1016/j.aei.2023.101988.
[6]	FU Yuanyuan, RAN Teng, XIAO Wendong, et al. GD-YOLO: an improved convolutional neural network architecture for real-time detection of smoking and phone use behaviors[J]. Digital Signal Processing, 2024, 151: DOI: 10.1016/j.dsp.2024.104554.
[7]	XIN Fangfang, HE Xinyu, YAO Caoxiu, et al. A real-time detection for miner behavior via DYS-YOLOv8n model[J]. Journal of Real-Time Image Processing, 2024, 21: DOI: 10.1007/s11554-024-01466-0.
[8]	罗国富, 王源, 李浩, 等. 基于改进YOLOv5s的智能车间工人不安全行为实时检测方法[J]. 计算机集成制造系统, 2024, 30(5):1610-1619.
	LUO Guofu, WANG Yuan, LI Hao, et al. Unsafe behavior real-time detection method of intelligent workshop workers based on improved YOLOv5s[J]. Computer Integrated Manufacturing Systems, 2024, 30(5): 1610-1619.
[9]	韩康, 李敬兆, 陶荣颖. 基于改进YOLOv7和ByteTrack的煤矿关键岗位人员不安全行为识别[J]. 工矿自动化, 2024, 50(3):82-91.
	HAN Kang, LI Jingzhao, TAO Rongying. Recognition of unsafe behaviors of key position personnel in coal mines based on improved YOLOv7 and ByteTrack[J]. Journal of Mine Automation, 2024, 50(3): 82-91.
[10]	徐浩钧, 胡啸峰, 吴建松. 化学实验室动火实验无人值守行为识别方法研究[J]. 中国安全生产科学技术, 2023, 19(12):135-141.
	XU Haojun, HU Xiaofeng, WU Jiansong. Research on identification method of unattended behavior in hot experiments of chemical laboratory[J]. Journal of Safety Science and Technology, 2023, 19(12): 135-141.
[11]	YU Weihao, SI Chenyang, ZHOU Pan, et al. MetaFormer baselines for vision[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024, 46(2): 896-912.
[12]	FENG Yifan, HUANG Jiangang, DU Shaoyi, et al. Hyper-YOLO: when visual object detection meets hypergraph computation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2025, 47(4): 2 388-2 401.
[13]	SUN Hang, WEN Yang, FENG Huijing, et al. Unsupervised bidirectional contrastive reconstruction and adaptive fine-grained channel attention networks for image dehazing[J]. Neural Networks, 2024, 176: DOI: 10.1016/j.neunet.2024.106314.
[14]	GRAHAM A. Kronecker products and matrix calculus with applications[M]. Horwood: Halsted Press, 1981: 21-35.
[15]	WITTEN I H, FRANK E, HALL M A, et al. Data mining: practical machine learning tools and techniques[M]. Burlington: Morgan Kaufmann, 2016: 181-186.

方案编号	防护面罩	手套	防护服	围巾
1	—	—	—	—
2	√	—	—	—
3	—	√	—	—
4	—	—	√
5	—	—	—	√
6	—	√	—	√
7	√	√	—	—
8	√	√	—	√
9	√		√	√
10	√	√	√	√

行为类型	识别目标	图像总数	标签总数
不安全着装	未佩戴防护面罩	2 085	1 843
	未穿着防护服	1 933	2 229
	戴手套作业	1 767	2 546
	戴围巾作业	1 789	1 836
违规操作	坐姿作业	4 161	4 295
违规操作	台钻无人值守	3 010	2 024

行为类型	P	R	mAP@0.5	F₁
未佩戴面罩	0.944 (0.941)	0.926 (0.923)	0.954 (0.953)	0.935 (0.932)
未穿防护服	1.000 (0.999)	0.987 (0.983)	0.995 (0.995)	0.993 (0.991)
戴手套作业	0.706 (0.735)	0.663 (0.682)	0.604 (0.651)	0.684 (0.708)
戴围巾作业	0.842 (0.880)	0.839 (0.826)	0.830 (0.841)	0.840 (0.852)
坐姿作业	0.882 (0.911)	0.868 (0.845)	0.952 (0.944)	0.875 (0.877)
无人值守	0.991 (0.982)	0.989 (0.989)	0.994 (0.99)	0.990 (0.985)

MANet	MetaFormer	AFGC Attention	P	R	mAP@0.5	F₁
—	—	—	0.706	0.663	0.604	0.684
√	—	—	0.687	0.663	0.606	0.675
—	√	—	0.693	0.682	0.645	0.687
—	—	√	0.734	0.654	0.661	0.692
√	√	—	0.731	0.668	0.629	0.698
√	—	√	0.718	0.605	0.621	0.657
—	√	√	0.657	0.67	0.596	0.663
√	√	√	0.735	0.682	0.651	0.708

模型	P	R	mAP@0.5	F₁
Faster R-CNN	0.278	0.356	0.202	0.312
SSD	0.628	0.427	0.405	0.508
YOLOv8x	0.710	0.644	0.633	0.675
YOLOv9e	0.693	0.632	0.644	0.661
YOLOv10x	0.660	0.621	0.621	0.640
YOLOv11x	0.706	0.663	0.604	0.684
YOLOv11x_MMA	0.735	0.682	0.651	0.708