摘要
为实现复杂背景下绝缘子快速、准确识别,提出了一种基于改进YOLOv3的绝缘子目标检测算法。采用K-means++算法调整anchor box,再改进Darknet-53特征提取网络,减少卷积层数和残差单元的使用次数来降低网络深度和运算量,然后增加一个尺度预测,将浅层特征信息加入到预测结果中以提升对小目标的检测敏感度。实验结果表明,改进的YOLOv3算法在绝缘子测试集上的表现良好,提高了检测速度并且准确率也有一定的提升。
In order to realize the rapid and accurate recognition of insulators under complex backgrounds, an improved insulator detection algorithm based on YOLOv3 is proposed. Use K-means++ algorithm to adjust the anchor box, and then improve the Darknet-53 feature extraction network, reduce the number of convolutional layers and the number of residual units to reduce the network depth and the amount of calculation, and then add a scale prediction to add shallow feature information into the prediction results to improve the detection sensitivity of small targets. The experimental results show that the improved YOLOv3 algorithm performs well on the insulator test set, which improves the detection speed and the accuracy rate.
作者
朱有产
郑怡
秦金磊
ZHU Youchan;ZHENG Yi;QIN Jinlei(School of Control and Computer Engineering,North China Electric Power University,Baoding 071003,China)
出处
《电瓷避雷器》
CAS
北大核心
2022年第3期166-171,共6页
Insulators and Surge Arresters
基金
中央高校基本科研业务费专项资金资助项目(编号:2018MS076,2020MS120)。