YOLO based Ocean Eddy Localization with AWS SageMaker

TL;DR

The paper addresses the challenge of localizing small-scale ocean eddies in SAR satellite imagery by deploying and benchmarking YOLOv5, YOLOv8, and YOLOv9 on AWS SageMaker, including single- and multi-GPU configurations. It combines a preprocessing pipeline with PCA-based restoration and augmentation, and uses SageMaker Ground Truth for labeling to enable scalable cloud-based training. Key findings show YOLOv9 variants offer strong localization performance, especially on challenging images, while smaller models provide efficient training and competitive accuracy; deployment challenges on SageMaker are documented to guide future cloud-Earth science workflows. The work demonstrates the practical feasibility of cloud-based eddy localization and highlights actionable limitations and opportunities to improve cloud ML pipelines for Earth observation data, with potential extensions to multi-satellite data and broader datasets.

Abstract

Ocean eddies play a significant role both on the sea surface and beneath it, contributing to the sustainability of marine life dependent on oceanic behaviors. Therefore, it is crucial to investigate ocean eddies to monitor changes in the Earth, particularly in the oceans, and their impact on climate. This study aims to pinpoint ocean eddies using AWS cloud services, specifically SageMaker. The primary objective is to detect small-scale (<20km) ocean eddies from satellite remote images and assess the feasibility of utilizing SageMaker, which offers tools for deploying AI applications. Moreover, this research not only explores the deployment of cloud-based services for remote sensing of Earth data but also evaluates several YOLO (You Only Look Once) models using single and multi-GPU-based services in the cloud. Furthermore, this study underscores the potential of these services, their limitations, challenges related to deployment and resource management, and their user-riendliness for Earth science projects.

Figures (9)

• Figure 1: Pipeline for Localizing Ocean Eddies using various YOLO models with SageMaker.
• Figure 2: An example of multiple Eddy containing SAR image converted using GDAL library.
• Figure 3: Image restoration process using PCA.
• Figure 4: Localization of Ocean Eddies: A Comparative Analysis of Various Models Applied to an Image.
• Figure 5: Localization of Ocean Eddies: Comparative Evaluation Across Multiple Models Using Additional Image.