%0 Journal Article
%T 基于YOLOv5的机场飞行区目标检测方法研究
Research on Target Detection Method of Airport Flight Area Based on YOLOv5
%A 张斗斗
%A 晏子奕
%A 周良杰
%A 蔡先乔
%A 汪海波
%A 惠康华
%J Modeling and Simulation
%P 432-441
%@ 2324-870X
%D 2022
%I Hans Publishing
%R 10.12677/MOS.2022.112040
%X 机场飞行区的飞机和车辆很多，人员活动复杂，需要检测的目标很多，而且需要实时检测目标等等。本文通过对四种主流目标检测算法的比较，找出了最适合机场飞行区的目标检测算法YOLOv5，以解决上述问题。利用YOLOv5x模型，对机场飞行区视频数据进行初步的目标检测和标注，然后手工修改和优化标注结果，组织数据集结构，合理分配训练集和测试集比例，构建适合机场飞行区的数据集。基于该数据集，本文采用深度学习算法YOLOv5训练出适用于机场飞行区的目标检测模型，实现了对机场飞行区主动目标的实时检测。通过在实验中动态调整算法策略，本文所训练的目标检测模型的效率达到了较高的水平。实验结果表明，改进后的算法具有95.5%的识别精度，满足机场飞行区的目标检测要求。
There are many aircraft and vehicles in the airport flight area, and the personnel activities are complicated, so there are a lot of targets to be detected, and they need to be detected in real time, etc. In this paper, we find out the most suitable target detection algorithm YOLOv5 for airport flight area to solve the above problems by comparing four mainstream target detection algorithms. Using the YOLOv5x model, we perform preliminary target detection and annotation on the airport flight area video data, and then manually modify and optimize the annotation results, organize the data set structure, reasonably allocate the training set and test set ratio, and construct a data set suitable for the airport flight area. Based on this dataset, this paper uses the deep learning algorithm YOLOv5 to train a target detection model applicable to the airport flight area and realize the real-time detection of active targets in the airport flight area. By dynamically adjusting the algorithm strategy in the experiments, the efficiency of the target detection model trained in this paper reaches a high level. The experimental results show that the improved algorithm has a recognition accuracy of 95.5%, which meets the target detection requirements of the airport flight area.
%K 机场飞行区，深度学习，YOLOv5，目标检测
Airport Flight Area
%K Deep Learning
%K YOLOv5
%K Target Detection
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=49577