MARIDA (Marine Debris Archive)

MARIDA (Marine Debris Archive) 是第一个基于多光谱 Sentinel-2 (S2) 卫星数据的数据集，它将海洋碎片与共存的各种海洋特征区分开来，包括马尾藻大型藻类、船舶、天然有机材料、波浪、尾流、泡沫、不同的水类型（即清澈、浑浊的水、富含沉积物的水、浅水）和云。 MARIDA 是一个开放访问数据集，使研究界能够探索某些漂浮物质、海况特征和水类型的光谱行为，开发和评估基于人工智能和深度学习架构的海洋碎片检测解决方案，以及卫星预处理管道。尽管它的设计目的是对多个机器学习任务有益，但它主要旨在对弱监督的像素级语义分割学习方法进行基准测试。 MARIDA 可以从存储库 Zenodo (https://doi.org/10.5281/zenodo.5151941) 下载。有关所有 ML 基准测试的快速入门指南和数据集的详细概述，请访问 https://marine-debris.github.io/。

MARIDA (Marine Debris Archive) is the first dataset based on multispectral Sentinel-2 (S2) satellite data, which distinguishes marine debris from various coexisting marine features, including Sargassum macroalgae, ships, natural organic materials, waves, wakes, foam, and different water types (i.e., clear water, turbid water, sediment-rich water, and shallow water) as well as clouds. MARIDA is an open-access dataset that enables the research community to explore the spectral behaviors of certain floating substances, sea state characteristics and water types, develop and evaluate marine debris detection solutions based on artificial intelligence and deep learning architectures, as well as satellite preprocessing pipelines. Although designed to benefit multiple machine learning tasks, its primary purpose is to benchmark weakly supervised pixel-level semantic segmentation learning methods. MARIDA can be downloaded from the Zenodo repository at https://doi.org/10.5281/zenodo.5151941. For a quick start guide to all ML benchmarks and a detailed overview of the dataset, please visit https://marine-debris.github.io.

AI搜集汇总

背景与挑战

背景概述

MARIDA (Marine Debris Archive) 是基于多光谱Sentinel-2卫星数据的首个海洋碎片检测数据集，能够区分海洋碎片与多种共存海洋特征（如马尾藻、船舶、天然有机材料等）。该数据集旨在支持人工智能和深度学习解决方案的开发与评估，特别是作为弱监督像素级语义分割的基准测试工具，并开放访问供研究社区使用。