为防止煤矿井下事故的发生，井下人员的识别定位起着重要作用，传统方法依赖于硬件设备进行识别定位，但其投入设备较多且没有与监控系统进行联系。近年来随着视频监控系统在井下重点区域全方位布置，基于监控系统的井下人员监测成为当前研究热点，由于目前无法通过视频监控呈现出人员身份信息，因此本文研究了基于深度学习的井下人员身份识别和基于射线法的井下危险区域人员入侵检测。主要工作内容如下：

（1）针对在井下昏暗环境中对人员身份识别度不高的问题，将反光号码牌贴在井下人员安全帽和工作服上，通过监控视频识别号码牌来实现识别人员身份，提出了一种基于YOLOv8-OCR的人员身份识别算法。在YOLOv8上下文感知的空间注意力机制中引入漏斗激活函数以提高井下环境中目标检测性能；在骨干网络添加通道空间注意力机制以解决号码牌尺寸小难以检测的问题；最后对检测到的反光号码牌区域用字符识别技术对区域内数字进行识别。在自建数据集上进行测试实验，结果表明识别准确率达到87.2％，单张图片识别速率用时为24.4ms，验证了改进模型的有效性和实时性。

（2）针对视频中的井下人员与危险区域的重合度无法反映目标真实位置问题，提出了一种将YOLOv8算法与射线法相结合的方法，从而实现对井下人员是否入侵危险区域进行判定。首先在骨干网络添加坐标注意力机制以提高对遮挡目标区域的检测精度；之后将YOLOv8的非极大值抑制改进为软非极大值抑制以解决目标漏检问题。在自建数据集上进行测试实验，结果表明平均精度均值达到88.7％，平均推理速度达到45FPS，有效地提高了井下危险区域入侵检测的精确性和可靠性。

（3）采用PyQt5设计了适用于井下人员监测的交互界面，将井下人员号码牌检测识别模块、危险区域入侵判定模块联系于一体。结合人员自身号码牌的对应，通过图形界面展示人员的身份信息及位置信息，并可方便地在危险区域入侵判定模块中，实时展示人员与所划分危险区域的位置关系。

In order to prevent accidents in coal mines, the identification and positioning of underground personnel plays an important role. The traditional method relies on hardware equipment for identification and positioning, but it requires a lot of equipment and is not connected with the monitoring system. In recent years, with the comprehensive deployment of video surveillance systems in key underground areas, underground personnel monitoring based on surveillance systems has become a current research hotspot. Since it is currently impossible to present personnel identity information through video surveillance, this article studies the identity of underground personnel based on deep learning. Identification and radiation-based detection of personnel intrusion in underground dangerous areas. The main work contents are as follows:

(1) Aiming at the problem of low personnel identification in the dim environment underground, reflective number plates are attached to the safety helmets and work clothes of underground personnel. The number plates are recognized by monitoring video to realize personnel identification. A personnel identification algorithm based on YOLOv8-OCR is proposed. The funnel activation function is introduced into the context-aware spatial attention mechanism of YOLOv8 to improve the target detection performance in the underground environment; a channel spatial attention mechanism is added to the backbone network to solve the problem of small size and difficulty in detecting number plates; finally, the detected reflective numbers are The card area uses character recognition technology to identify the numbers in the area. Test experiments were conducted on a self-built data set, and the results showed that the recognition accuracy reached 87.2%, and the single image recognition rate took 24.4ms, which verified the effectiveness and real-time performance of the improved model.

(2) Aiming at the problem that the overlap between underground personnel and dangerous areas in the video cannot reflect the real position of the target, a method combining YOLOv8 algorithm with ray method is proposed to determine whether underground personnel have invaded dangerous areas. First, a coordinate attention mechanism is added to the backbone network to improve the detection accuracy of occluded target areas; then the non-maximum suppression of YOLOv8 is improved to soft non-maximum suppression to solve the problem of target miss detection. Test experiments were conducted on a self-built data set, and the results showed that the average accuracy reached 88.7% and the average inference speed reached 45FPS, which effectively improved the accuracy and reliability of intrusion detection in underground dangerous areas.

(3) An interactive interface suitable for underground personnel monitoring was designed using PyQt5. Integrate the underground personnel number plate detection and identification module and the dangerous area intrusion determination module into one. Combined with the corresponding number plate of the person, the identity information and location information of the person are displayed through the graphical interface, and the positional relationship between the person and the divided dangerous area can be displayed in real time in the dangerous area intrusion determination module.

[1] 李洪言, 张景谦, 陈健斌, 等. 2021年全球能源转型面临挑战——基于《bp世界能源统计年鉴(2022)》[J]. 天然气与石油, 2022, 40(06): 129-138.

[2] 王世栋. 煤矿企业安全管理十八法[J]. 中国煤炭, 2022, 48(S1): 158-163.

[3] 李晓理, 张博, 王康, 等. 人工智能的发展及应用[J]. 北京工业大学学报, 2020, 46(06): 583-590.

[4] El-Gayar M M, Soliman H. A comparative study of image low level feature extraction algorithms[J]. Egyptian Informatics Journal, 2013, 14(2): 175-181.

[5] Huang S Y, An W J, Zhang D S, et al. Image classification and adversarial robustness analysis based on hybrid quantum–classical convolutional neural network[J]. Optics Communications, 2023, 533: 129287.

[6] Tian Y. Artificial intelligence image recognition method based on convolutional neural network algorithm[J]. IEEE Access, 2020, 8: 125731-125744.

[7] Zhou D X. Deep distributed convolutional neural networks: Universality[J]. Analysis and applications, 2018, 16(06): 895-919.

[8] Gaba S, Budhiraja I, Kumar V, et al. A federated calibration scheme for convolutional neural networks: Models, applications and challenges[J]. Computer Communications, 2022, 192: 144-162.

[9] Wang J, Zeng X, Duan S, et al. Image Target Recognition Based on Improved Convolutional Neural Network[J]. Mathematical Problems in Engineering, 2022.

[10] Zhang J, Meng Y, Chen Z. A small target detection method based on deep learning with considerate feature and effectively expanded sample size[J]. IEEE Access, 2021, 9: 96559-96572.

[11] He K, Gkioxari G, Dollár P, et al. Mask r-cnn[C]//Proceedings of the IEEE international conference on computer vision. 2017: 2961-2969.

[12] 王鹏. 煤矿安全监控系统智能化现状及发展[J]. 能源与节能, 2023(2): 153-155.

[13] 王栋. 人工智能OCR技术的应用研究[J]. 电子技术与软件工程, 2022, 219(01): 122-125.

[14] Van den Broeck W A J, Goedemé T. Combining Deep Semantic Edge and Object Segmentation for Large-Scale Roof-Part Polygon Extraction from Ultrahigh-Resolution Aerial Imagery[J]. Remote Sensing, 2022, 14(19): 4722.

[15] 荣耀, 安晓宇. 智能化开采中视频信息的应用现状及展望[J]. 煤炭科学技术,2021,49(S1):119-123.

[16] 李现国, 李斌, 刘宗鹏, 等. 井下视频行人检测方法[J]. 工矿自动化, 2020, 46(02): 54-58.

[17] Cui Lizhen, Zhang Qingyu, Guo Qianqian, et al. Underground personnel behavior pattern recognition model based on CNN-LSTM [J]. Radio Engineering, 2023, 53(06): 1375-1381.

[18] 付燕, 窦晓熠, 叶鸥. 基于YOLO的井下人员速度测量方法研究[J]. 煤炭工程, 2022, 54(02): 160-165.

[19] Li X, Wang S, Liu B, et al. Improved YOLOv4 network using infrared images for personnel detection in coal mines[J]. Journal of Electronic Imaging, 2022, 31(1): 013017-013017.

[20] Imam M, Baïna K, Tabii Y, et al. Anti-Collision System for Accident Prevention in Underground Mines using Computer Vision[C]//Proceedings of the 6th International Conference on Advances in Artificial Intelligence. 2022: 94-101.

[21] Wang L, Li W, Zhang Y, et al. Pedestrian detection based on YOLOv2 with skip structure in underground coal mine[C]//2017 IEEE 3rd Information Technology and Mechatronics Engineering Conference (ITOEC). IEEE, 2017: 1216-1220.

[22] Wu F, Liu W, Wang S, et al. Improved mine pedestrian detection algorithm based on YOLOv4-Tiny[C]//Third International Symposium on Computer Engineering and Intelligent Communications (ISCEIC 2022). SPIE, 2023, 12462: 644-649.

[23] 谢斌红, 袁帅, 龚大立. 基于RDB-YOLOv4的煤矿井下有遮挡行人检测[J]. 计算机工程与应用, 2022, 58(05): 200-207.

[24] 范伟强, 王雪瑾, 张颖慧, 等. 改进YOLOv7和DeepSORT的井下人员检测与跟踪算法[J]. 煤炭科学技术, 2024: 1-13.

[25] 董观利, 宋春林. 基于视频的矿井行人越界检测系统[J]. 工矿自动化, 2017, 43(02): 29-34.

[26] Wang J. Research and implementation of intrusion detection algorithm in video surveillance[C]//2016 International Conference on Audio, Language and Image Processing (ICALIP). IEEE, 2016: 345-348.

[27] Tran S V T, Lee D, Bao Q L, et al. A Human Detection Approach for Intrusion in Hazardous Areas Using 4D-BIM-Based Spatial-Temporal Analysis and Computer Vision[J]. Buildings, 2023, 13(9): 2313.

[28] Zhao J, Xu Y, Zhu W, et al. Real-Time Early Safety Warning for Personnel Intrusion Behavior on Construction Sites Using a CNN Model[J]. Buildings, 2023, 13(9): 2206.

[29] Wang W, Wang S, Guo Y, et al. Obstacle detection method of unmanned electric locomotive in coal mine based on YOLOv3-4L[J]. Journal of Electronic Imaging, 2022, 31(2): 023032-023032.

[30] 李纳森. 基于YOLO的井下目标检测方法[D]. 徐州: 中国矿业大学, 2021.

[31] 刘浩, 刘海滨, 孙宇, 等. 煤矿井下员工不安全行为智能识别系统[J]. 煤炭学报, 2021, 46(S2): 1159-1169.

[32] Liu B, Tian B, Qiao J. Mine track obstacle detection method based on information fusion[C]//Journal of Physics: Conference Series. IOP Publishing, 2022, 2229(1): 012023.

[33] Mei X, Zhou X, Xu F, et al. Human intrusion detection in static hazardous areas at construction sites: Deep learning–based method[J]. Journal of Construction Engineering and Management, 2023, 149(1): 04022142.

[34] Liu H, Duan X, Chen H, et al. DBF‐YOLO: UAV Small Targets Detection Based on Shallow Feature Fusion[J]. IEEJ Transactions on Electrical and Electronic Engineering, 2023, 18(4): 605-612.

[35] Liu H, Sun F, Gu J, et al. Sf-yolov5: A lightweight small object detection algorithm based on improved feature fusion mode[J]. Sensors, 2022, 22(15): 5817.

[36] Lou H, Duan X, Guo J, et al. DC-YOLOv8: small-size object detection algorithm based on camera sensor[J]. Electronics, 2023, 12(10): 2323.

[37] 陈航, 邱晓晖. 基于卷积神经网络和池化算法的表情识别研究[J]. 计算机技术与发展, 2019, 29(01): 61-65.

[38] 龙曼仪, 李茂军, 张辉, 等. 基于轻量型卷积神经网络的交通标志识别[J]. 计算技术与自动化, 2020, 39(04): 112-118.

[39] Liu J G, Yang X J, Feng Y Y, et al. New fractional derivative with sigmoid function as the kernel and its models[J]. Chinese Journal of Physics, 2020, 68: 533-541.

[40] 张焕, 张庆, 于纪言. 卷积神经网络中激活函数的性质分析与改进[J]. 计算机仿真, 2022, 39(04): 328-334.

[41] Redmon J, Divvala S, Girshick R, et al. You only look once: Unified, real-time object detection[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 779-788.

[42] Liu W, Anguelov D, Erhan D, et al. Ssd: Single shot multibox detector[C]//Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016, Proceedings, Part I 14. Springer International Publishing, 2016: 21-37.

[43] Redmon J, Farhadi A. YOLO9000: better, faster, stronger[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2017: 7263-7271.

[44] Redmon J, Farhadi A. Yolov3: An incremental improvement[J]. arXiv preprint, 2018, arXiv:1804.02767.

[45] He K, Zhang X, Ren S, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 770-778.

[46] Bochkovskiy A, Wang C Y, Liao H Y M. Yolov4: Optimal speed and accuracy of object detection[J]. arXiv preprint, 2020, arXiv:2004.10934.

[47] Li C, Li L, Jiang H, et al. YOLOv6: A single-stage object detection framework for industrial applications[J]. arXiv preprint, 2022, arXiv:2209.02976.

[48] Wang C Y, Bochkovskiy A, Liao H Y M. YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2023: 7464-7475.

[49] Sirisha U, Praveen S P, Srinivasu P N, et al. Statistical analysis of design aspects of various YOLO-based deep learning models for object detection[J]. International Journal of Computational Intelligence Systems, 2023, 16(1): 126.

[50] Wang C Y, Liao H Y M, Wu Y H, et al. CSPNet: A new backbone that can enhance learning capability of CNN[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops. 2020: 390-391.

[51] LIN T Y, DOLLÁR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]//Proceedings of The IEEE Conference on Computer Vision and Pattern Recognition, 2017: 936-944.

[52] LIU S, QI L, QIN H, et al. Path aggregation network for instance segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018: 8759-8768.

[53] 周燕, 韦勤彬, 廖俊玮, 等. 自然场景文本检测与端到端识别: 深度学习方法[J]. 计算机科学与探索, 2023, 17(03): 577-594.

[54] Watanabe S, Hori T, Kim S, et al. Hybrid CTC/attention architecture for end-to-end speech recognition[J]. IEEE Journal of Selected Topics in Signal Processing, 2017, 11(8): 1240-1253.

[55] He P, Huang W, Qiao Y, et al. Reading scene text in deep convolutional sequences[C]//Proceedings of the AAAI conference on artificial intelligence. 2016, 30(1).

[56] Shi B, Bai X, Yao C. An end-to-end trainable neural network for image-based sequence recognition and its application to scene text recognition[J]. IEEE transactions on pattern analysis and machine intelligence, 2016, 39(11): 2298-2304.

[57] 苟先太, 陶明江, 李欣, 等. 基于宽接收域的实时人体姿态估计网络程与设计[J]. 计算机工程与设计, 2023, 44(01): 247-254.

[58] Ma N, Zhang X, Sun J. Funnel activation for visual recognition[C]//Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XI 16. Springer International Publishing, 2020: 351-368.

[59] Woo S, Park J, Lee J Y, et al. Cbam: Convolutional block attention module[C]//Proceedings of the European conference on computer vision (ECCV). 2018: 3-19.

[60] 王珂, 杨芳, 姜杉. 光学字符识别综述[J]. 计算机应用研究, 2020, 37(S2): 22-24.

[61] 杜紫薇, 周恒, 李承阳, 李忠博, 谢永强, 董昱辰, 齐锦. 面向深度卷积神经网络的小目标检测算法综述[J]. 计算机科学, 2022, 49(12): 205-218.

[62] Liu J, Dai P, Han G, et al. Combined CNN/RNN video privacy protection evaluation method for monitoring home scene violence[J]. Computers and Electrical Engineering, 2023, 106: 108614.

[63] Priyadarsini K, Janahan S K, Thirumal S, et al. Recognition of the Multioriented Text Based on Deep Learning[M]//Expert Clouds and Applications: Proceedings of ICOECA 2023. Singapore: Springer Nature Singapore, 2023: 367-375.

[64] JaidedTeam. Easy OCR[DB/OL]. https://github.com/JaidedAI/EasyOCR, 2021-11-05.

[65] Xuan W, Jian-She G, Bo-Jie H, et al. A lightweight modified YOLOX network using coordinate attention mechanism for PCB surface defect detection[J]. IEEE Sensors Journal, 2022, 22(21): 20910-20920.

[66] Abadi M. TensorFlow: learning functions at scale[C]//Proceedings of the 21st ACM SIGPLAN international conference on functional programming. 2016: 1-1.

[67] Jia Y, Shelhamer E, Donahue J, et al. Caffe: Convolutional architecture for fast feature embedding[C]//Proceedings of the 22nd ACM international conference on Multimedia. 2014: 675-678.

[68] 王晓辉, 聂小华, 常亮. 基于Qt的专用有限元软件GUI模块的设计与开发[J]. 计算机应用与软件, 2020, 37(01): 21-26+65.