基于改进YOLOv5的输电线路瓷瓶检测

聂嘉琦

国家电网银川供电公司，宁夏银川，750004；

摘要：搭载了目标检测器的无人机在进行输电线路瓷瓶检测时。受制于边缘设备内存限制，小参数量的目标检测模型无法保证较高的检测精度，导致在实际工业应用中的效果差强人意。针对传统YOLOv5目标检测模型定位精度不佳的问题，本文提出改进的YOLOv5算法，重构了YOLOv5的输出头部，输出概率分布表示的坐标。并进一步引入定位分布焦点损失，提升了定位精度。实验数据证明了本文所提出改进算法切实有效地提升了检测精度。

关键词：YOLOv5；输电线路瓷瓶；目标检测

参考文献

[1]宋思齐.高压电气设备红外检测技术研究[D].武汉:华中科技大学,2016.

[2]Dun Nan Liu Rui Hou, Wen Zhuo Wu, Jing Wen Hua , Xuan Yuan Wang, Bo Pang Research on infrared image enhancement and segmentation of power equipment based on partial differential equation[J]. Journal of Visual Communication and Image Representation.2019.64.

[3]X. Jianbo, K. Chen, C. Tian, F. Xiaohua and 2. Longwu, "Study on the Intelligent Analysis Method of the Infrared Detection of Electrical Equipment," 2018 3rd International Conference on Smart City and Systems Engineering (ICSCSE)Xiamen, China, 2018, pp. 500-504.

[4]吴俊杰. 无人机智能巡视系统在变电站巡检中的应用 [J]. 农村电气化, 2023, (10): 21-23+88. DOI:10.13882/j.cnki.ncdqh.2023.10.006.

[5]He K, Zhang X, Ren S, et al. Spatial pyramid pooling in deep convolutional networks for visual recognition[J]. IEEE transactions on pattern analysis and machine intelligence, 2015, 37(9):1904-1916.

[6]Lin T Y, Dollar P, Girshick R, et al. Feature pyramid networks for object detection[C]//Proceedings of the IEEE conference on computer vision and pattern recognition.2017: 2117-2125.

[7]Liu S, QiL, Qin H, et al. Path aggregation network for instance segmentation[C]J//Proceedings of the IEEE conference on computer vision and pattern recognition. 2018: 8759-8768.

[8]LI X,WANG W,WU L, et al. Generalized focal loss: Learning qualified and distributed bounding boxes for dense object detection[A]. 2020.