| 基于CenterNet 的航空遥感图像目标检测 |
| |
| 引用本文： | 杨曦中,高冠鸿,熊智,张玲.基于CenterNet 的航空遥感图像目标检测[J].航空电子技术,2022,53(1):1-7. |
| |
| 作者姓名： | 杨曦中 高冠鸿 熊智 张玲 |
| |
| 作者单位： | 航空电子系统综合技术重点实验室，上海200233；;南京航空航天大学自动化学院，南京,211106 |
| |
| 基金项目： | 军科委技术领域基金资助(2020-JCJQ-JJ-153) |
| |
| 摘 要： | 为实现高精度的航空图像目标检测，将Anchor free 的目标检测算法CenterNet 应用到检测中，同时
使用Resnet50 主干网络，并引入CIoU 损失替代原有损失函数对网络模型做出了改进。改进后的算法在RSOD
与DIOR 数据集上进行测试，结果显示在保证网络轻量化的前提下检测精度有明显的提高，证明了算法在航空
目标检测方面的可行性与准确性。
|
| 关 键 词： | 神经网络 目标检测 深度学习 CenterNet |
| 收稿时间： | 2021/9/23 0:00:00 |
| 修稿时间： | 2021/11/3 0:00:00 |
Object Detection on Remote Sensing Image Using CenterNet |
| |
| YANG Xi-zhong,GAO Guan-hong,XIONG Zhi,ZHANG Ling.Object Detection on Remote Sensing Image Using CenterNet[J].Avionics Technology,2022,53(1):1-7. |
| |
| Authors: | YANG Xi-zhong GAO Guan-hong XIONG Zhi ZHANG Ling |
| |
| Affiliation: | Science Technology on Avionics Integration Laboratory, Shanghai 200233, China;;College of Automation Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China |
| |
| Abstract: | In order to achieve high-precision aerial image object detection, an Anchor free object detection algorithm
CenterNet is applied to the detection. Meanwhile, the Resnet50 backbone is implemented and the CIoU loss is introduced
to replaces the original loss function to improve the network model. The improved algorithm is tested on the
RSOD and DIOR dataset, and the results show that the detection accuracy has been significantly improved under the
premise of ensuring the lightweight of the network, which proved the feasibility and accuracy of the algorithm in aviation
target detection. |
| |
| Keywords: | convolutional neural network object detection deep learning CenterNet |
|
| 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》下载全文 |
