基于改进YOLOv8的飞机检测研究

Study on Aircraft Inspection Based on Improved YOLOv8

随着无人机技术和航空监控系统的迅速发展,高效准确的飞机检测技术变得日益重要。提出了一种基于改进的YOLOv8模型的飞机检测方法,旨在提高检测的准确性和实时性。首先,引入了反向残差注意力模块(iRMB),通过改进的注意力机制增强模型对飞机特征的学习能力。其次,采用中心化特征金字塔(EVC)模块,优化了特征提取过程,增强了模型对不同尺度飞机的检测能力。此外还采用了改进的距离交并比(MDIoU)作为损失函数,进一步提升了模型的定位精度。在公开数据集Caltech101的飞机类别上的实验结果表明,与现有的飞机检测方法相比,提出的方法在检测精度和召回率方面分别达到98.2%和98.9%,特别是在复杂背景和多尺度目标检测场景中表现更为突出。该研究的成果对于提高航空安全监控系统的效能具有重要意义。

This paper proposes an aircraft detection method based on the improved YOLOv8 model,aiming to improve the accuracy and real-time performance of detection.First,this paper introduces the inverse residual attention module(iRMB)to enhance the model's ability to learn aircraft features through an improved attention mechanism.Second,the centered feature pyramid(EVC)module is used to optimize the feature extraction process and enhance the model's ability to detect aircraft at different scales.In addition,an improved distance intersection and merger ratio(MDIoU)is adopted as the loss function in this paper to further enhance the localization accuracy of the model.Experimental results on the aircraft category of the publicly available dataset Caltech101 show that compared with existing aircraft detection methods,the method proposed in this paper achieves 98.2%and 98.9%in terms of detection precision and recall,respectively,and performs more prominently especially in complex background and multi-scale target detection scenarios.