首页|一种基于CAL-YOLO的海上航空搜救目标检测算法

一种基于CAL-YOLO的海上航空搜救目标检测算法

扫码查看
针对在海上复杂环境搜救任务下,目标尺寸小,且搜救飞机在不同高度和视角下观测到的搜救目标特征各异或存在遮挡等因素造成搜救目标漏检、误检的问题,提出了一种基于CAL-YOLO的海上航空搜救目标检测算法.该算法基于YOLOv8s开源框架改进,引入坐标注意力机制和小目标检测层,以优化海上航空搜救目标检测性能,从而提高海上航空搜救工作效率.实验结果表明,该算法在mAP50的评价指标上提高了11.8%,在mAP50:95评价指标上提高了8.8%.
CAL-YOLO-based target detection algorithm for maritime aviation search and rescue
In order to address the issue of missed and falsely detected search and rescue targets in complex maritime environments,where targets are small in size and exhibit varying characteristics or there are occlusions when observed by search and rescue aircraft at different heights and angles,a CAL-YOLO-based maritime avia-tion search and rescue target detection algorithm is proposed.This algorithm is improved based upon the YO-LOv8s open-source framework,introducing the coordinate attention(CA)mechanism and small target detection layer to optimize the performance of maritime aviation search and rescue target detection,thereby enhancing the efficiency of maritime aviation search and rescue operations.The experimental results show that the algorithm im-proves the evaluation index of mAP50 by 11.8%and mAP50:95 by 8.8%.

maritime aviation search and rescueYOLOv8s algorithmsmall target detection layercoordi-nate attention(CA)mechanism

郭秀文、杨林、刘济民、张朝阳

展开 >

92728部队,上海 200436

海上航空搜救 YOLOv8s算法 小目标检测层 坐标注意力机制

2024

空天预警研究学报
空军预警学院

空天预警研究学报

影响因子:0.39
ISSN:2097-180X
年,卷(期):2024.38(3)
  • 10