基于深度学习的电力巡检目标检测与追踪研究

Research on Power Inspection Target Detection and Tracking Based on Deep Learning

随着我国电力行业的飞速发展,传统的人工电力巡检方式已无法满足当前行业的发展需求。文章提出一种基于深度学习的电力巡检目标检测与追踪模型。该模型通过在YOLOv7中引入CBAM(Convolutional Block Attention Module)注意力模块,构建了CBAM-YOLOv7改进检测算法,并将其识别结果作为DeepSORT(Simple Online and Realtime Tracking With A Deep Association Metric)目标追踪算法的输入,实现了对电网故障的有效检测与追踪。实验结果表明,相较于原YOLOv7算法,改进后的CBAM-YOLOv7算法在精确度、召回率、平均精度3个评价指标上均有提升,而DeepSORT算法的平均MOTA值也达到87.817%。这证明了该模型能够在真实复杂场景下准确地定位电网故障。

With the rapid development of the power industry in China,traditional manual power inspection methods can no longer meet the current industry demands.This paper proposes a deep learning-based model for object detection and tracking in power inspection.The model introduces the CBAM(Convolutional Block Attention Module)attention mechanism into YOLOv7,creating an improved detection algorithm called CBAM-YOLOv7.The detection results from this model are used as input for the DeepSORT(Simple Online and Realtime Tracking With A Deep Association Metric)tracking algorithm,achieving effective detection and tracking of power grid faults.Experimental results indicate that compared to the original YOLOv7 algorithm,the improved CBAM-YOLOv7 algorithm shows enhancements in precision,recall,and mean average precision metrics,while the DeepSORT algorithm achieves an average MOTA value of 87.817%.It is proved that the model can accurately locate power grid faults in real complex scenarios.