Abstract
Accurate analysis and identification of underground monitoring videos in coal mines can prevent safety accidents caused by unsafe behaviors of underground personnel and protect their safety. In light of this context, this study enhances the traditional YOLOv11 algorithm for target detection and proposes a fast and effective method for identifying unsafe behaviors among underground personnel in the complex environment of coal mines. Firstly, a statistical analysis of the most common types of unsafe behaviors in current underground coal mines is conducted, exploring the classification of miners’ unsafe behaviors into item-type, action-type, and area-type categories. Secondly, based on the characteristics of these unsafe behaviors, we propose dataset augmentation and denoising preprocessing techniques to enhance fine-grained feature extraction. Simultaneously, we introduce the parameter-free SimAM to improve the saliency mapping of miners’ behaviors. Finally, we optimize the YOLOv11 algorithm by incorporating a function enhancement module and the K-means + + anchor frame, and we propose a dual-model recognition method for target detection that integrates the YOLOv11 algorithm with the YOLOv11-Pose algorithm. To validate the performance of our non-standard miners’ behavior recognition method, we test it on a self-constructed dataset. The research results demonstrate that our method can quickly and effectively recognize unsafe behaviors among underground personnel. Compared to traditional methods, our approach significantly improves recognition accuracy on both the self-constructed dataset and the public dataset, achieving a mean Average Precision (mAP) of 95.7%, an accuracy rate of 95.3%, and a recall rate of 95.1%. These findings are significant for preventing underground safety accidents.
Data availability
No datasets were generated or analysed during the current study.
References
Cao, X. et al. Unsafe mining behavior identification method based on an improved ST-GCN. Sustainability 15 (2), 1041 (2023).
Yao, W. et al. Study on the recognition of coal miners’ unsafe behavior and status in the hoist cage based on machine vision. Sensors 23 (21), 8794 (2023).
Lijuan, L. et al. YOLOv5-SFE: An algorithm fusing spatio-temporal features for detecting and recognizing workers’ operating behaviors. Adv. Eng. Inform. 56, 101988 (2023).
Liang, F. Construction site safety helmet wearing detection method based on improved YOLOv5. J. Phys. Conf. Ser. 2560 (1), 012042 (2023).
Ren, S. et al. Faster R-CNN: Towards real-time object detection with region proposal networks. IEEE Trans. Pattern Anal. Mach. Intell. 39 (6), 1137–1149 (2016).
Khan, M. N., Das, S. & Liu, J. Predicting pedestrian-involved crash severity using inception-v3 deep learning model. Accid. Anal. Prev. 197, 107457 (2024).
Kolar, Z., Chen, H. & Luo, X. Transfer learning and deep convolutional neural networks for safety guardrail detection in 2D images. Autom. Constr. 89, 58–70 (2018).
Zhe, C. et al. OpenPose: Realtime multi-person 2D pose estimation using part affinity fields. IEEE Trans. Pattern Anal. Mach. Intell. 43 (1), 172–186 (2019).
Vukicevic, A. M. et al. Deep learning-based recognition of unsafe acts in manufacturing industry. IEEE Access. 11, 103406–103418 (2023).
Srivastava, N., Mansimov, E. & Salakhudinov, R. Unsupervised learning of video representations using lstms. Int. Conf. Mach. Learn. PMLR, 843–852 (2015).
Guo, Y. et al. Wearable sensor data based human behavior recognition: a method of data feature extraction. J. Comput.-Aided Des. Comput. Graph. 33 (8), 1246–1253 (2021).
Liu, Y. et al. Real-time human activity recognition based on time-domain features of multi-sensor. Zhongguo Guanxing Jishu Xuebao (Journal Chin. Inert. Technol.). 25 (4), 455–460 (2017).
Gao, H. et al. Healthcare system from multisensor collaboration and human action recognition. Sens. Mater. 36 (2024).
Yang, X. et al. Indoor activity and vital sign monitoring for moving people with multiple radar data fusion. Remote Sens. 13 (18), 3791 (2021).
Chen, C. et al. Study on human hazardous behavior recognition and monitoring system in slide facilities based on improved HRNet network. Int. J. Adv. Comput. Sci. Appl., 16(3). (2025).
Lee, T. & Mihailidis, A. An intelligent emergency response system: preliminary development and testing of automated fall detection. J. Telemed. Telecare. 11 (4), 194–198 (2005).
Özyer, T., Ak, D. S. & Alhajj, R. Human action recognition approaches with video datasets—A survey. Knowl. Based Syst. 222, 106995 (2021).
Yan, S., Xiong, Y. & Lin, D. Spatial temporal graph convolutional networks for skeleton-based action recognition. Proc. AAAI Conf. Artif. Intell. 32(1) (2018).
Aslanyan, M. On mobile pose estimation and action recognition design and implementation. Pattern Recognit. Image Anal. 34 (1), 126–136 (2024).
Wang, C., Zhang, H. & Zhai, Z. Real time dangerous action warning system based on graph convolution neural network. Acad. J. Comput. Inform. Sci. 5 (6), 89–94 (2022).
Standard for Classification of Casualties. In Enterprise Employees (GB 6441-1986) (State Bureau of Standards, 1986).
Bishop, C. M. & Nasrabadi, N. M. Pattern Recognition and Machine Learning (Springer, 2006).
Redmon, J. & Farhadi, A. Yolov3: An incremental improvement. arXiv preprint arXiv:1804.02767 (2018).
Graves, A. & Graves, A. Long short-term memory. Supervised sequence labelling with recurrent neural networks. 37–45 (2012).
Mnih, V. et al. Human-level control through deep reinforcement learning. Nature 518 (7540), 529–533 (2015).
Funding
This work was supported and financed from the Langfang Science and Technology Program (Grant number: 2024013001), the Hebei Natural Science Foundation (Grant number: E2023508021), the Fundamental Research Funds for the Central Universities (Grant number: 3142021002, 3142024005).
Author information
Authors and Affiliations
Contributions
Conceptualization, J.L. and Q.Z.; Data curation, J.L., Q.Z. and D.J.; Formal analysis, J.L., Q.Z. and D.J.; Funding acquisition, Q.Z. and Y.L.; Investigation, J.L., Q.Z. and S.C.; Methodology, J.L., Q.Z., D.J. and Y.L.; Project administration, J.L. and Q.Z.; Resources, S.C., Y.L. and Q.Z.; Software, Q.Z., D.J. and Y.H.; Supervision, J.L. and Q.Z.; Validation, D.J., Y.H. and S.C.; Visualization, Q.Z., D.J. and Y.H.; Writing-original draft, J.L., Q.Z. and D.J.; Writing-review and editing, Q.Z. and D.J.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Juan, L., Zhu, Q., Jiang, D. et al. Research on identification method and application of unsafe behavior of coal mine personnel. Sci Rep (2026). https://doi.org/10.1038/s41598-026-47077-6
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-026-47077-6
