现代电子技术2025，Vol.48Issue(11)：114-120,7.DOI:10.16652/j.issn.1004-373x.2025.11.018
融合小尺寸动态蛇形卷积的太阳能电池板检测YOLO模型
YOLO model of solar panel inspection fused with small-sized dynamic serpentine convolution
摘要
Abstract
In view of the low detection accuracy in the detection task of solar panel surface defects,a YOLO model of solar panel detection fused with small-sized dynamic serpentine convolution is proposed.The small-sized dynamic serpentine convolution that can adapt to the bundle-like defects is mainly designed,which enhances the model's ability to characterize crack defects.An efficient channel attention(ECA)mechanism is added to the neck of the model,and the weighted gradient features are fused to enhance the extraction ability of key features.The simulation experiments were based on the open data set.The experimental results show that the mAP@0.5 of the proposed method is 88.1%,which is 5%,15.8%,and 1.6%higher than that of YOLOv7,Faster R-CNN and YOLOX-S,respectively.To sum up,the proposed method can more accurately complete the task of detecting surface defects of solar cells.关键词
太阳能电池板/缺陷检测/梯度信息/动态蛇形卷积/高效通道注意力机制/YOLO模型Key words
solar panel/defect detection/gradient information/dynamic serpentine convolution/efficient channel attention mechanism/YOLO model分类
电子信息工程引用本文复制引用
汪宇玲,常佳熠..融合小尺寸动态蛇形卷积的太阳能电池板检测YOLO模型[J].现代电子技术,2025,48(11):114-120,7.基金项目
国家自然科学基金项目(62066003) （62066003）
国家留学基金项目(CSC202208360143) （CSC202208360143）
