| 基于改进Yolov3 算法的舰船目标检测识别系统 |
| |
| 引用本文： | 唐崇武,刘洪喜,代长安.基于改进Yolov3 算法的舰船目标检测识别系统[J].航空电子技术,2022,53(2):39-46. |
| |
| 作者姓名： | 唐崇武 刘洪喜 代长安 |
| |
| 作者单位： | 中国航空无线电电子研究所，上海 200233;空军装备部驻上海地区第二军事代表室，上海 200233 |
| |
| 摘 要： | 针对复杂战场环境下对海目标检测识别的需求，设计了一种基于改进Yolov3 算法的海面舰船目标实
时检测识别系统。使用微调分类网络、增加训练尺度、聚类目标边框维度、二级特征分类等方法对Yolov3 检
测识别网络模型进行了优化，在提高识别精度的同时有效降低了漏检率和虚警率。实验结果表明，优化后的网
络模型在自建的舰船图像数据库中将检测识别平均准确率提高到了79.3%，对真实海上航拍视频中舰船目标识
别的平均准确率达到了81% 以上。
|
| 关 键 词： | 舰船目标检测识别 卷积神经网络 深度特征 模型训练 目标聚类 |
| 收稿时间： | 2021/11/1 0:00:00 |
| 修稿时间： | 2022/2/24 0:00:00 |
Improved Yolov3 Based Ship Target Detection and Recognition System |
| |
| TANG Chong-wu,LIU Hong-xi,DAI Chang-an.Improved Yolov3 Based Ship Target Detection and Recognition System[J].Avionics Technology,2022,53(2):39-46. |
| |
| Authors: | TANG Chong-wu LIU Hong-xi DAI Chang-an |
| |
| Institution: | China National Aeronautical Radio Electronics Research Institute, Shanghai 200233;;The 2nd Military Representative Office of Air Force Equipment Department in Shanghai Area, Shanghai 200233 |
| |
| Abstract: | For the needs of sea target detection and recognition in complex battle field environment, an improved
Yolov3-algorithm-based real-time detection and recognition system is proposed for marine ship targets. Several methods
are utilized to optimize the detection network model of the system, including network fine-tuning, training scales
increasing, target box clustering and two-phase feature classification. These optimizations can effectively reduce the
missed detection rate and false alarm rate. Experimental results show that the designed system raises the accuracy of
detection and recognition to 79.3% on ship image database built, and achieves the mean accuracy on real aerial videos
of marine ship scenes over 81%. |
| |
| Keywords: | ship target recognition convolutional neural networks deep features model training target clustering |
|
| 点击此处可从《航空电子技术》浏览原始摘要信息 |
| 点击此处可从《航空电子技术》下载免费的PDF全文 |
