基于YOLO算法的海洋中尺度涡三维结构构建

3D Structure Construction of Mesoscale Eddies Based on YOLO Intelligent Algorithm

摘要:

海洋中尺度涡是一种常见的中尺度海洋现象,研究海洋中尺度涡的分布及运动特性对航运、气候、军事等具有重要作用,海洋中尺度涡的识别是海洋学和计算机科学领域的一个热门研究课题。本文运用深度学习的方法和框架,对中尺度涡的二维识别和三维结构构建展开研究分析。首先,获取全球海洋再分析数据并进行流线可视化,构建涡旋流线数据集；其次,利用YOLO v5s卷积神经网络对涡旋流线数据集进行训练,并对南海区域中尺度涡进行有效检测。实验结果表明,YOLO v5s训练后得到最优模型经过测试,平均检测精度均值达到了86.10%；最后,根据涡旋检测结果,对检测出的同时刻不同深度的涡旋判断是否属于同一涡旋,确定后进行该涡旋的三维结构构建。

Abstract:

Marine mesoscale eddies are a common marine mesoscale phenomenon. Studying the distribution and motion characteristics of mesoscale eddies plays an important role in shipping, climate, and military affairs. At present, the identification of marine mesoscale eddies is a hot research topic in the field of oceanography and computer science. In this paper, the identification and three-dimensional structure construction of mesoscale eddies were studied and analyzed by using the method and framework of deep learning. Firstly, the global marine physics reanalysis data were obtained to visualize streamlines to build the dataset of eddies streamlines. Secondly, the YOLO v5s convolutional neural network was used to train the eddies streamline dataset and successfully detect mesoscale eddies in the South China Sea region. The experimental results showed that the optimal model obtained after YOLO v5s training had been tested, and the mean precision accuracy had reached 86.10%, which could provide an effective basis for the identification of mesoscale eddies and the construction of 3D structures. Finally, based on the eddy detection results at the same time, the detected eddies at different depths were determined whether they belonged to the same eddy, then the three-dimensional structure of the eddy was built.