GLH-Water_全球大幅面卫星影像水体检测数据集

为解决相关数据集的不足并提升地表水分割模型的泛化能力，我们首次推出GLH-water数据集。该数据集包含250幅超高分辨率（地面采样距离GSD=0.3米）卫星影像，每幅影像尺寸为12,800×12,800像素。这些影像采集自全球不同地理位置，并包含人工标注信息。
GLH-water数据集具有以下五项显著优势：样本尺寸大，涵盖海量样本，样本地理覆盖范围广，影像采集时间跨度大，包含多样化的地表水景观类型。

To address the shortcomings of existing relevant datasets and improve the generalization ability of surface water segmentation models, we propose the GLH-water dataset for the first time. This dataset contains 250 ultra-high-resolution satellite images, each with a resolution of 12,800 × 12,800 pixels and a ground sampling distance (GSD) of 0.3 meters. These images are collected from various geographical locations across the globe, and all are paired with manually annotated labels.

The GLH-water dataset features five prominent advantages: large individual sample size, a substantial volume of samples, wide geographical coverage, a long time span of image acquisition, and diverse types of surface water landscapes.

搜集汇总

背景与挑战

背景概述

GLH-Water是一个全球大幅面卫星影像水体检测数据集，包含250幅超高分辨率（0.3米）的卫星图像，每幅尺寸为12,800 × 12,800像素，覆盖2011年至2022年全球多个地点，并附带人工标注的水体标签。该数据集具有样本尺寸大、数量多、地理覆盖广、时间跨度长和景观类型多样等特点，旨在增强分割模型在全球地表水检测中的泛化能力，适用于遥感和水体检测研究。