基于卷积神经网络的航母舰桥识别

Aircraft Carrier Bridge Recognition Based on Convolutional Neural Network

针对反舰导弹攻击航母时雷达关机的问题,提出转变制导方式为可见光制导,并采用YOLOv5神经网络对舰桥自动识别的技术方案。由于航母数据集较少,采用Mosaic数据增强的方法,不仅增加了数据集数量,而且提高了网络的鲁棒性。将特征明显、目标较大的舰桥作为主要识别对象。在对数据集进行建立之后,完成了对其的测试以及训练,从而获得了以下研究结论:在不同情况、不同角度的航母识别中,检测准确率可达到90%以上,对航母准确识别跟踪任务具有重要意义。

The change of guidance method to visible light guidance is proposed for the problem of radar shutdown during antiship missile attack on aircraft carriers.A technical solution is adopted for automatic identification of the bridge using YOLOv5 neural network.Mosaic data augmentation is used due to the small carrier data set.This not only increases the number of data sets,but also improves the robustness of the network.The bridge with obvious features and larger targets is used as the main recognition object.Through the creation,training and testing of the dataset,the final experimental results are that the detection accuracy can reach more than 90%in the identification of aircraft carriers in different situations and angles,which is significant for the accurate identification and tracking task of aircraft carriers.