基于YOLOv7的直升机航巡电塔目标检测算法研究

为了提升直升机航巡的智能化水平,解决人工作业效率低下的问题,提出了一种基于YOLOv7的直升机航巡电塔目标检测算法。首先,建立直升机航巡电塔数据集,将其分为训练集和测试集。然后,构建YOLOv7目标检测网络模型,并利用训练样本库进行聚类分析,获取目标候选区域的先验尺寸,使用随机梯度下降迭代地修改网络参数,最终实现直升机航巡电塔目标检测。实验结果表明,该方法精度较高、时效性较好,准确度达到94.9%,召回率达到90.5%,算法时间消耗仅39.5 ms,满足直升机航巡电塔目标检测的需求。

Pylon Object Detection of Helicopter Cruise Based on YOLOv7

Object detection algorithms based on deep learning have become increasingly mature,making it possible to promote the intellectualization of helicopter navigation patrol while solving the inefficiency problems of manual inspection.This study proposes an object detection method based on YOLOv7 for helicopter navigation pylon through the establishment of a helicopter cruise pylon dataset that is divided into training and test set.By constructing the YOLOv7 object detection network model and determining the prior size of the pylon candidate region through clustering analysis,the network parameters are modified using stochastic gradient descent iteration to ultimately achieve successful helicopter cruise pylon object detection.Experimental results reveal that the proposed method has high accuracy and good timeliness,with detection accuracy of 94.9% and recall rate of 90.3%,and algorithm running time of only 39.5 ms,meeting the requirements of helicopter pylon detection.