基于改进YOLOv3_iny的压敏电阻表面缺陷检测

Surface defect detection of varistor based on improved YOLOv3;iny

针对当前目标检测算法网络复杂,对平台设备要求高;而轻量化网络YOLOv3_iny对压敏电阻表面缺陷的检测精度较低,容易出现漏检错检的情况,提出了基于YOLOv3_iny的改进算法DAYOLOv3_iny。DAYOLOv3_iny构建了深度可分离卷积块替代标准卷积,使用卷积操作进行下采样,使检测网络在减少网络参数量的同时增加了特征的提取;并在网络中引入了通道注意力模块和空间注意力模块,增强了检测网络对重要特征信息的学习。在自制的压敏电阻表面缺陷数据集上进行实验,结果表明,DAYOLOv3_iny的mAP值为92.23%,较改进前提升了12.25%;改进后的DAYOLOv3_iny模型大小为YOLOv3_iny的55.42%,仅18.9 MB。实验表明,DAYOLOv3_iny对压敏电阻表面缺陷的检测精度较高,能够有效改善漏检错检的情况,且网络模型较小,对硬件平台要求不高,易于在性能受限的平台部署。

In view of the complexity of the current target detection algorithm network and the high demand on the platform equipment;And the low detection accuracy of the lightweight network YOLOv3_iny on the varistor surface defects,which is prone to miss and wrong detection,an improved algorithm DAYOLOv3_iny based on YOLOv3_iny is proposed.DAYOLOv3_iny constructs the deep separable convolution block to replace the standard convolution,use the convolution operation to downsample,so that the detection network can reduce the number of network parameters and increase the feature extraction;The channel attention module and spatial attention module are introduced into the network to enhance the learning of important feature information.The results on a self-made varistor surface defect data set show that the MAP value of DAYOLOv3_iny is 92.23%,which is 12.25% higher than before.The size of the improved DAYOLOv3_iny model is 55.42%of that of YOLOv3_iny,which is only 18.9 MB.Experiments show that DAYOLOv3_iny has a higher detection accuracy for varistor surface defects,and can effectively improve the situation of missed and wrong detection.Besides,the network model is small,and it does not require high hardware platform,so it is easy to deploy in the platform with limited performance.