基于改进YOLOv4-tiny的输电线路目标识别算法

对输电线路周围的典型目标进行检测对于防止输电线路外部破坏有着重要意义。传统目标检测方法没有针对输电线路周围目标尺度变化大、小目标多等进行有效设计，存在识别速度慢、容易误报漏报等问题。基于YOLOv4-tiny目标检测模型的基本框架，提出了一种改进的YOLOv4-tiny目标检测模型来检测输电线路周围的典型目标。在原先YOLOv4-tiny网络的骨干网上额外引出了一层特征层以提取更多的特征；在原特征金字塔网络结构的基础上引入空洞空间卷积池化金字塔模块，使得模型能在3种不同尺度的特征图上提取更多的特征；同时为解决检测过程中正负样本数量不均衡问题，使用Focal损失函数代替二分交叉熵损失函数。实验结果表明，在牺牲较少检测速度的情形下，模型精度提升了9.92%。

Transmission Line Target Recognition Algorithm Based on Improved YOLOv4-tiny

It is important to detect the typical targets around the transmission lines to prevent the external damage of the transmission lines.The traditional target detection method does not design effectively for the scale change of the target around the transmission line,and there are problems such as slow recognition speed,easy to false alarms.Based on the basic framework of YOLOv4-tiny object detection model,an improved YOLOv4-tiny object detection model is proposed to detect typical targets around transmission lines.An additional feature layer is introduced on the backbone of the original YOLOv4-tiny to extract more features.An atrous spatial pyramid pooling module is introduced on the basis of the original feature pyramid network structure,so that the model can extract more features on the feature map of three different scales.At the same time,in order to solve the problem of imbalance in the number of positive and negative samples in the detection process,the Focal loss function is used instead of the binary cross-entropy loss function.The experimental results show that the accuracy of the model is improved by 9.92% at the expense of less detection speed.