CNN based Ocean Eddy Detection Using Cloud Services
CNN based Ocean Eddy Detection Using Cloud Services
复制标题
DOI：
10.1109/igarss52108.2023.10283367
复制
发表时间：
2023-07
期刊：
影响因子：
--
通讯作者：
S. A. Mostafa;Jinbo Wang;Benjamin Holt;Sanjay Purushotham;Jianwu Wang
中科院分区：
文献类型：
--
作者：
S. A. Mostafa;Jinbo Wang;Benjamin Holt;Sanjay Purushotham;Jianwu Wang
这项研究的重点是使用部署在AWS云平台（特别是SageMaker和EC2）上的卷积神经网络（CNN）在卫星遥感图像中进行小规模海洋涡流（<20 km）检测。我们的目标是简化工作流程，使气候变化领域的服务更容易获得。在这项工作中，我们提出了一个基于CNN的海洋涡流检测模型，并使用Sage-Maker和EC2进行部署。通过对数据集应用主成分分析（PCA），我们提出的方法分别实现了超过95%和94%的训练和验证准确率。我们在部署服务时考虑了可用性、性能和成本比较。我们的分析表明，SageMaker和EC2非常有能力构建基于CNN的服务，尽管在部署服务方面存在一些挑战和与资源相关的限制。
This study focuses on small-scale ocean eddy (<20km) detection in satellite remote images using Convolutional Neural Network (CNN) deployed on AWS cloud platforms, specifically SageMaker and EC2. Our goal is to streamline the workflow and make the services accessible in the climate change domain. In this work, we proposed a CNN-based Ocean Eddy detection model and it is deployed using Sage-Maker and EC2. Our proposed approach achieved more than 95% and 94% training and validation accuracy respectively by applying principal component analysis (PCA) on the dataset. We considered the usability, performance, and cost comparison while deploying the services. Our analysis shows that SageMaker and EC2 are highly capable of building CNN-based services, though there are some challenges in deploying services and limitations related to resources.