摘要
Abstract
To address the issues of low efficiency and poor real-time performance in the manual supervision of safety equipment(helmets and life jackets)during marine survey operations,this study aims to achieve intelligent safety monitoring.This paper proposes an object detection method based on YOLO11.A dedicated dataset for marine surveys was constructed,containing four target categories:person,helmet,work life jacket,and life jacket.Using this dataset,YOLOv5,YOLOv8,and YOLO11 models were trained and tested,with a comprehensive comparative analysis conducted from the perspectives of detection accuracy,speed,and computational complexity.Experimental results show that the YOLO11 model achieved an Precision(P)of 83.1%and a mean Average Precision(mAP)of 82.2%on the validation set,operating at a speed of 416.7 frames per second(FPS).Compared to YOLOv8,YOLO11 achieved higher accuracy while reducing the number of parameters by approximately 15%.The YOLO11 model achieves a good balance between accuracy and speed.Its lightweight characteristics and high detection accuracy meet the requirements for real-time detection of safety equipment usage in marine environments,providing an effective technical reference for the intelligent monitoring of safety in marine operations.关键词
深度学习/YOLO11/海洋调查/安全帽检测/救生衣检测/目标检测Key words
deep learning/YOLO11/marine survey/helmet detection/life jacket detection/object detection分类
信息技术与安全科学
