基于YOLOv10n-knot的电力安全带挂点检测算法

郑艳梅; 张欣; 芦碧波; 苏浩

doi:10.13637/j.issn.1009-6094.2025.085810.13637/j.issn.1009-6094.2025.0858

Chinese

您当前的位置：

首页 >

文章列表页 >

基于YOLOv10n-knot的电力安全带挂点检测算法

更新时间：2026-04-17

- 基于YOLOv10n-knot的电力安全带挂点检测算法
- Vol. 26, Issue 4, Pages: 1316-1327(2026)
- 作者机构：
  
  河南理工大学计算机科学与技术学院
- 作者简介：
- 基金信息：
- DOI：10.13637/j.issn.1009-6094.2025.085810.13637/j.issn.1009-6094.2025.0858
  CLC： TP391.41;TM08
- Published：2026
- 稿件说明：
移动端阅览
[1]郑艳梅,张欣,芦碧波,等.基于YOLOv10n-knot的电力安全带挂点检测算法[J].安全与环境学报,2026,26(04):1316-1327.
[1]郑艳梅,张欣,芦碧波,等.基于YOLOv10n-knot的电力安全带挂点检测算法[J].安全与环境学报,2026,26(04):1316-1327. DOI： 10.13637/j.issn.1009-6094.2025.085810.13637/j.issn.1009-6094.2025.0858.

DOI：10.13637/j.issn.1009-6094.2025.0858.

摘要

针对电力安全工作规程中高空作业安全带“高挂低用”的强制安全要求，提出了一种安全带挂点检测模型YOLOv10n-knot。首先，针对挂点目标尺寸小、特征易被主干网络忽略的问题，在主干网络中引入大核可分离卷积注意力机制，以增强安全带挂点的特征提取能力；其次，为提高对挂点的识别能力，在头部网络增加小目标检测层；最后，为增强对图像的特征提取能力，在颈部模块集成动态语义感知上采样算子，以提升模型对复杂场景的适应能力。试验表明，改进模型挂点检测的均值平均精度达98.5%

参数量减少至2.60×10

。与YOLOv8n、RT-DETR等模型相比，本模型在准确率、召回率及均值平均精度指标上均有显著提升。

Abstract

To comply with the mandatory safety requirements of the “high-position and low-use” standard in high-altitude electric power operations

this paper presents an enhanced high-precision safety harness anchor point detection model named YOLOv10n-knot. This model assesses compliance by detecting the relative spatial position between the safety harness and its anchor point. To tackle the challenge of detecting small anchor point targets that may be overlooked by the backbone network

we introduce a large-kernel separable convolution attention mechanism into the backbone architecture. This module utilizes sequential 1D convolutions in both horizontal and vertical directions

significantly improving the extraction of fine-grained local features. Secondly

a P2 small-object detection head with a resolution of 160×160 is added to the

model's architecture. This addition enhances the fusion of shallow high-resolution features with deep semantic information

thereby reducing the loss of critical features that can occur due to excessive downsampling. Thirdly

the model incorporates a dynamic semantic-aware upsampling operator in the neck region. By employing a content-aware mechanism to generate adaptive convolution kernels

this module enhances the quality of cross-scale feature fusion

thereby improving the model's robustness and localization precision in complex backgrounds. Experimental results demonstrate that the YOLOv10n-knot model excels in the anchor point detection task

achieving an accuracy of 97.4%

a recall of 97.3%

an mAP@50 of 98.5%

and an mAP@50:95 of 77.9%

all with a lightweight parameter size of only 2.60×10

. Compared with mainstream models such as YOLOv8n and RT-DETR

the proposed model shows improvements of 3.2 percentage points and 3.8 percentage points in accuracy

8.8 percentage points and 4.3 percentage points in recall

and 14.8 percentage points and 13.2 percentage points in mAP@50:95

respectively. Additionally

inference speed and memory consumption were evaluated across various hardware platforms. The results indicate that YOLOv10n-knot exhibits fast inference times and low memory usage

making it suitable for real-time deployment. In summary

YOLOv10n-knot achieves high detection accuracy while maintaining a lightweight structure. It demonstrates strong generalization and practical value

offering an effective solution for intelligent compliance detection of safety harness usage in high-altitude electric power operations.

关键词

Keywords

references

辛保安，单葆国，李琼慧，等.“双碳”目标下“能源三要素”再思考[J].中国电机工程学报，2022,42(9):3117-3125.XIN B A,SHAN B G,LI Q H,et al.Rethinking of the “three elements of energy” toward carbon peak and carbon neutrality[J].Proceedings of the CSEE,2022,42(9):3117-3125.

FANG W L,DING L Y,LUO H B,et al.Falls from heights:a computer vision-based approach for safety harness detection[J].Automation in Construction,2018,91(1):53-61.

卫潮冰.基于CNN的安全智能监测识别算法[J].电子设计工程，2020,28(4):64-68.WEI C B.Intelligent security monitoring and recognition algorithms based on CNN[J].Electronic Design Engineering,2020,28(4):64-68.

潘志敏，王梓糠，蒋毅，等.基于深度迁移学习的电力作业安全带佩戴检测[J].计算机仿真，2022,39(5):95-101.PAN Z M,WANG Z K,JIANG Y,et al.A deep transfer learning-based approach for power safety harness detection[J].Computer Simulation,2022,39(5):95-101.

WU L F,CAI N,LIU Z J,et al.A one-stage deep learning framework for automatic detection of safety harnesses in high-altitude operations[J].Signal,Image and Video Processing,2023,17(1):75-82.

石彦鹏，潘作为，成浩天.基于YOLOv5和少样本学习的高空作业安全带检测[J].电子器件，2024,47(1):215-220.SHI Y P,PAN Z W,CHENG H T.Safety belt detection for aerial work based on YOLOv5 and few-shot learning[J].Chinese Journal of Electron Devices,2024,47(1):215-220.

LI S Y,LIU X Y.A study on detection algorithm of safety apparatus wearing by workers at heights based on deep learing[J].International Journal of Innovative Computing,Information and Control,2023,19(5):1593-1603.

XU Z J,HUANG J J,HUANG K.A novel computer vision-based approach for monitoring safety harness use in construction[J].IET Image Processing,2023,17(4):1071-1085.

李华，吴立舟，薛曦澄，等.基于计算机视觉的高处临边作业安全巡检[J].中国安全科学学报，2023,33(9):69-75.LI H,WU L Z,XUE X C,et al.Computer vision based safety inspection of high abutting edges[J].China Safety Science Journal,2023,33(9):69-75.

李永福，陈立斌，惠君伟，等.基于EPSA-YOLOv5电力高空作业安全带佩戴检测[J].西安工程大学学报，2024,38(2):18-25.LI Y F,CHEN L B,HUI J W,et al.Safety belt wearing detection for electric aloft work based on EPSA-YOLOv5[J].Journal of Xi'an Polytechnic University,2024,38(2):18-25.

朱建宝，桑顺，邓伟超，等.可提高检测精度与速度的电力安全带智能检测方法[J].电气自动化，2025,47(1):20-22.ZHU J B,SANG S,DENG W C,et al.Intelligent detection method for power safety belts improving detection accuracy and speed[J].Electrical Automation,2025,47(1):20-22.

段在鹏，翁培辉，蓝志宇，等.电网登高违章作业轻量级智能检测模型[J].安全与环境学报，2025,25(1):175-183.DUAN Z P,WENG P H,LAN Z Y,et al.Lightweight intelligent detection model for high-altitude violation operations in electric power grids[J].Journal of Safety and Environment,2025,25(1):175-183.

宋维.基于空间和通道双重注意力机制的变电站施工安全带分割方法研究[D].保定：华北电力大学，2020.SONG W.Research on substation construction safety belt segmentation method based on double attention mechanism of space and channel[D].Baoding:North China Electric Power University,2020.

钱照国，苟瀚文，康立烨，等.基于人体关键点的安全带违规佩戴检测[J].计算机仿真，2024,41(1):211-216.QIAN Z G,GOU H W,KANG L Y,et al.Detection of illegal wearing of seat belts based on key points of the human body[J].Computer Simulation,2024,41(1):211-216.

张历，许逵，李鑫卓，等.基于空间和通道注意力机制的高空作业场景安全带语义分割算法分析[J].集成电路应用，2025,42(1):394-396.ZHANG L,XU K,LI X Z,et al.Analysis of semantic segmentation algorithm for safety belts in high-altitude work scenarios based on spatial and channel attention mechanism[J].Application of IC,2025,42(1):394-396.

ZHOU Q,LIU D.Improving safety in high-altitude work:semantic segmentation of safety harnesses with CEMFormer[J].Symmetry,2024,16(11):1449.

孙慧颖，许永刚，戴超，等.全过程高挂主动安全型防坠落装置的研制[J].电力安全技术，2015,17(1):65-68.SUN H Y,XU Y G,DAI C,et al.Development of full-process high-hanging active safety anti-fall device[J].Electric Safety Technology,2015,17(1):65-68.

BIAN D,TANG M W,LING M G,et al.A refined methodology for small object detection:multi-scale feature extraction and cross-stage feature fusion network[J].Digital Signal Processing,2025,164:105297.

梁燕，何孝武，邵凯，等.改进YOLOv8的无人机航拍图像目标检测算法[J].计算机工程与应用，2025,61(1):121-130.LIANG Y,HE X W,SHAO K,et al.Target detection algorithm for UAV images based on improved YOLOv8[J].Computer Engineering and Applications,2025,61(1):121-130.

PENG B,XIONG S D,WANG Y J,et al.MI-DETR:a small object detection model for mixed scenes[J].Displays,2025,88:103052.

LAU K W,PO L M,REHMAN Y A U.Large separable kernel attention:Rethinking the large kernel attention design in CNN[J].Expert Systems with Applications,2024,236:121352.

WANG J Q,CHEN K,XU R,et al.CARAFE:content-aware reassembly of features[C]//Proceedings of the 2019 IEEE/CVF International Conference on Gomputer Vision (ICCV).Piscataway,N.J.:Institute of Electrical and Electronics Engineers,2019:3007-3016.

Views

238

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

复杂场景下融合多尺度特征与注意力机制的安全帽检测

园区环境下的烟头小目标检测算法研究

融合多尺度特征的煤矿井下粉尘检测方法

考虑时空特征的瓦斯涌出量预测模型及其可解释性分析

改进粒子群算法优化CNNLSTMAttention模型在安全生产事故预测中的应用

Related Author

邱云飞

齐焓玮

金海波

白亚腾

刘润翌

陈万志

白佳慧

韩雪峰

Related Institution

辽宁工程技术大学软件学院

中国中煤能源集团有限公司

中国矿业大学人工智能研究院

辽宁工程技术大学鄂尔多斯研究院

中国矿业大学(北京)应急管理与安全工程学院

AI问答

Postal code：100079
Tel：（010）53879206 Email：tmw@bjxintong.com.cn
Technical support is provided by Beijing Founder electronics co., LTD 京ICP备09082226号-64 京公网安备11010602201714号
It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.

⁰