Replication Data for: An Efficient Solution for Semantic Segmentation of Three Ground-based Cloud datasets

The SWIMSEG dataset, which includes 1013 cloud images, is divided into three parts according to the common practice: training set 911, testing set 51, validation set 51. The original dataset can be downloaded from http://vintage.winklerbros.net/swimseg.html. The BENCHMARK dataset includes 32 cloud images patches and corresponding cloud masks.SWIMSEG 数据集（SWIMSEG Dataset）共包含1013幅云图像，按照行业通用惯例划分为三个子集：训练集911张、测试集51张、验证集51张。该数据集的原始版本可通过 http://vintage.winklerbros.net/swimseg.html 下载。BENCHMARK 数据集（BENCHMARK Dataset）包含32幅云图像块及其对应的云掩膜。

搜集汇总

背景与挑战

背景概述

该数据集是用于复现地面云图像语义分割研究的，包含SWIMSEG和BENCHMARK两个子数据集。SWIMSEG数据集有1013张云图像，已按常规划分为训练集（911张）、测试集（51张）和验证集（51张）；BENCHMARK数据集则包含32张云图像块及其对应的云掩码。数据集主题为地球与环境科学，采用CC0 1.0许可证，便于公开使用和共享。