改进YOLOv8的建筑物毁伤效果评估算法
Improved YOLOv8 Algorithm for Building Damage Effect Evaluation
针对现代战争中战场环境复杂和传统评估方法过于依赖主观经验的问题,提出一种改进YOLOv8的建筑物毁伤效果评估算法。首先,在样本输入端采用Mosaic-9进行图像预处理,提升网络模型的泛化能力。其次,在骨干网络中引入高效多尺度注意力机制(EMA),同时利用新构造的快速跨阶段局部网络融合模块(FC2f)模块,提升网络模型的特征提取能力和运行效率。最后,采用Scylla交并比损失函数(SIoU Loss)对网络损失函数进行优化,进一步提升网络模型的检测评估精度和运行效率。实验结果表明,该算法检测评估精度较高,运算速度较快,具有一定的军事应用价值。
Aiming at the problems of complex battlefield environment and excessive reliance on subjective experience in traditional evaluation methods in modern war, an improved YOLOv8 algorithm for building damage effect evaluation is proposed. Firstly, the Mosaic-9 is used for image preprocessing at the sample input to improve the generalization ability of the network model. Secondly, the efficient multi-scale attention (EMA) mechanism is introduced into the backbone network, and the newly constructed fast cross stage partial network fusion (FC2f) is used to improve the feature extraction ability and operation efficiency of the network model. Finally, the Scylla intersection over union loss (SIoU Loss) function is used to optimize the network loss function to further improve the detection and evaluation accuracy and operation efficiency of the network model. Experimental results show that this algorithm has high detection and evaluation accuracy, fast operational speed, and certain military application value.
改进YOLOv8 / 建筑物 / 毁伤评估 / 图像预处理 / 损失函数优化
improved YOLOv8 / buildings / damage assessment / image preprocessing / loss function optimization
| [1] |
武青平,李高宇.无人化智能化战争形态下的作战体系建设问题思考[J].军事文摘,2021(11):31-34. |
| [2] |
黄林江,于小红,王杰娟,解析美军马赛克战概念的内涵和战场变化的关系[J].信息工程大学学报,2023,24(5):627-633. |
| [3] |
杨青青,樊桂花.基于高分辨率遥感图像的建筑物毁伤效果评估[J].电子设计工程.2018,26(21):6-10. |
| [4] |
魏鑫,李晓婷.基于模糊推理的综合毁伤效果评估方法[J]. 智能计算机与应用,2022,12(7):146-150. |
| [5] |
杨延平.基于图像变化检测的毁伤效果评估技术研究[D].成都:电子科技大学,2013:59-60. |
| [6] |
张宗腾,张琳,谢春燕,基于改进GA-BP神经网络的目标毁伤效果评估[J].火力与指挥控制,2021,46(11):43-48. |
| [7] |
Ultralytics. Explore Ultralytics YOLOv8[DB/OL]. (2023-01-10)[2024-08-05]. |
| [8] |
周觐,高岚岚,刘巍.智能态势认知关键需求分析[J]. 指挥控制与仿真,2024,46(2):8-17. |
| [9] |
王杰,张上,张岳,改进YOLOv5的军事飞机检测算法[J].无线电工程,2024,54(3):589-596. |
| [10] |
曾志超,徐玥,王景玉,基于SOE-YOLO轻量化的水面目标检测算法[J].图学学报,2024,1(3):72-80. |
| [11] |
单慧琳,王硕洋,童俊毅,增强小目标特征的多尺度光学遥感图像目标检测[J].光学学报,2024,44(6):382-394. |
| [12] |
徐艺博,颜佳润,曾志文,无通信条件下基于视觉毁伤评估的弹群对地目标自主攻击决策[J].兵工学报,2024,32(5):1-14. |
| [13] |
李杰,李勇斌,郑娄,基于YOLO-OpenMax的水声通信信号开集识别方法[J].信息工程大学学报,2024,25(3):258-264. |
| [14] |
秦振,李学伟,刘宏哲.基于改进SSD的鲁棒小目标检测算法[J].东北师大学报(自然科学版),2023,55(4):59-66. |
| [15] |
崔鹏,杨海峰,蔡江辉,王玉鹏.多尺度局部聚类的Kmeans-DETR目标检测方法[J].小型微型计算机系统,2024,45(5):1136-1142. |
| [16] |
|
武警部队军事理论课题(WJJY24JL0141)
/
| 〈 |
|
〉 |