首页|基于STM 32的输电线路缺陷检测方法研究

基于STM 32的输电线路缺陷检测方法研究

扫码查看
基于STM32单片机无线传输系统设计,同时结合蓝牙模块、传感器模块传送获取输电线路的图像数据,并且针对输电线路图像背景单一、数据集数量少的特点,提出一种基于改进YOLOv7 的输电线路缺陷检测方法.本文对数据集进行自动扩充利用少量图片建立合适的数据集,主干网络替换为ConvNeXt主干网络,并引入将ECA注意力机制,将PANet替换为BiFPN进行特征融合.利用消融实验和对比实验验证改进算法的有效性和优越性,结果表明改进算法的mAP由 89.62%上升到 92.27%,对背景单一场景具有更好的检测性能目标.
A Method for Transmission Line Defect Detection Based on STM32
Based on STM32 microcontroller wireless transmission system design,while combining the Bluetooth module,sensor module transmission,the image data of the transmission line are obtained.Targeting at the characteristics of the single image background and poor quantity of data sets of transmission lines,a defect-detecting method for transmission lines grounded on improved YOLOv7 is devised.Firstly,automatic data enrichment is used to solve the problem of small sample data.Secondly,the backbone network is replaced by the ConvNeXt backbone network,and the ECA is introduced into the ConvNeXt backbone network.A new algorithm that uses BiFPN for fea-ture fusion is proposed,replacing PANet.Ablation and comparative experiments are conducted to verify the effectiveness and superiority of the devised algorithm.The consequences demonstrate that the improved algorithm achieves an mAP of 92.27%,which is 2.65%higher than that of the YOLOv7 algorithm.Furthermore,the proposed algorithm exhibits better detection performance for single background scenes.

STM32YOLOv7transmission linesdefect detectionECA

黄学达、陈思思、袁刘湘、何鑫、魏钰颖

展开 >

重庆邮电大学自动化学院,重庆 400065

STM32 YOLOv7 输电线路 缺陷检测 注意力机制

2024

电子器件
东南大学

电子器件

CSTPCD
影响因子:0.569
ISSN:1005-9490
年,卷(期):2024.47(6)