[1]贺强*,刘洋,游鑫,等.基于改进YOLOv9的电力设备红外目标检测模型[J].计算机技术与发展,2025,(07):182-189.[doi:10.20165/j.cnki.ISSN1673-629X.2025.0112]
HE Qiang*,LIU Yang,YOU Xin,et al.Target Detection Model for Infrared Image of Power Equipment Based on Improved YOLOv9[J].,2025,(07):182-189.[doi:10.20165/j.cnki.ISSN1673-629X.2025.0112]
点击复制
基于改进YOLOv9的电力设备红外目标检测模型(
)
《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]
- 卷:
-
- 期数:
-
2025年07期
- 页码:
-
182-189
- 栏目:
-
新型计算应用系统
- 出版日期:
-
2025-07-10
文章信息/Info
- Title:
-
Target Detection Model for Infrared Image of Power Equipment Based on Improved YOLOv9
- 文章编号:
-
1673-629X(2025)07-0182-08
- 作者:
-
贺强1*; 刘洋1; 游鑫1; 张瑞亮1; 王玉峰1; 黄谦1; 何雨非2
-
1. 中国南方电网有限责任公司超高压输电公司贵阳局,贵州 贵阳 550000;
2. 华北电力大学(保定) 电子与通信工程系,河北 保定 071003
- Author(s):
-
HE Qiang1*; LIU Yang1; YOU Xin1; ZHANG Rui-liang1; WANG Yu-feng1; HUANG Qian1; HE Yu-fei2
-
1. Guiyang Bureau of China Southern Power Grid Co. ,Ltd. ,Ultra High Voltage Transmission Company,Guiyang 550000,China;
2. Department of Electronic and Communication Engineering,North China Electric Power University (Baoding),Baoding 071003,China
-
- 关键词:
-
电力设备; 红外图像检测; YOLOv9算法; 注意力机制; 轻量化模型; 损失函数
- Keywords:
-
power equipment; infrared target detection; YOLOv9; attention mechanism; light-weight model; loss function
- 分类号:
-
TP183
- DOI:
-
10.20165/j.cnki.ISSN1673-629X.2025.0112
- 摘要:
-
在电力巡检领域,电力设备红外图像检测的重要性不言而喻,其准确性直接关系到电力系统的稳定运行与安全性。 针对当前复杂场景下红外图像目标检测存在的识别精度不足、分类混淆以及漏检误检等问题,该文提出了一种基于 YOLOv9 的改进检测方法。 该方法首先通过引入空间到深度卷积层(SPD-Conv),实现了模型主干网络的轻量化设计,有效提升了计算效率与实时性。 随后,结合卷积块注意力模块(CBAM),进一步优化了模型对关键特征的提取能力,增强了检测的准确性。 此外,还采用 Focal-IoU 作为损失函数,强化了模型的判别能力,减少了定位误差。 实验结果表明,在南方电网电力设备红外数据集上,该算法取得了显著的成效,mAP 值达到了94. 4% ,同时推理速度也达到了59. 5 f/ s,充分展示了该方法在检测精度与计算效率方面的双重优势。 与现有方法相比,该模型在各类电力设备上的检测能力均有所提升,为电力巡检领域提供了一种更为可靠、高效的解决方案。
- Abstract:
-
In the field of power inspection,the importance of infrared image detection of power equipment is self-evident,and its accuracy is directly related to the stable operation and safety of the power system. In order to solve the problems of insufficient recognition accuracy,classification confusion and missed detection and false detection in infrared image target detection in complex scenes, an improved detection method based on YOLOv9 is proposed. Firstly,by introducing the space-to-deep convolutional layer (SPD-Conv),the lightweight design of the model backbone network is realized,which effectively improves the computational efficiency and real-time performance. Subsequently,combined with the Convolutional Block Attention Module ( CBAM), the model’s ability to extract key features was further optimized,and the accuracy of detection was enhanced. In addition,Focal-IoU is used as the loss function,which strengthens the discriminant ability of the model and reduces the positioning error. The experimental results show that the proposed algorithm has achieved remarkable results on the infrared dataset of power equipment of China Southern Power Grid,with the mAP value reaching 94. 4% and the inference speed reaching 59. 5 f/ s,which fully demonstrates the dual advantages of the proposed method in terms of detection accuracy and computational efficiency. Compared with the existing methods,the detection ability of the proposed model on various power equipment has been improved,which provides a more reliable and efficient solution for the field of power inspection.
更新日期/Last Update:
2025-07-10