Classification Data set : Land Cover Classification with Gaussian Processes using spatio-spectro-temporal features

Classification data set (train, validation, test) from the study area based on 27 tiles on the south of the France. Data set are provided for each eco-climatic region. The size corresponds to the data set DS-A. Only one random pixel sampling is provided: seed 0. This data set was used to train Gaussian Processes, Random Forest, Multilayer Perceptron and Lightweight Temporal Self-Attention models. For further details see section VI-A-1 of the pre-print article "Land Cover Classification with Gaussian Processes using spatio-spectro-temporal features ". This article is available here. The implementation of the models is available in the open source repository.本数据集为基于法国南部27幅瓦片（tiles）的研究区域分类数据集，包含训练集、验证集与测试集。数据集按各生态气候区分别提供，其规模与数据集DS-A一致，仅提供一种随机像素采样方案：随机种子为0。本数据集曾用于训练高斯过程（Gaussian Processes）、随机森林（Random Forest）、多层感知机（Multilayer Perceptron）以及轻量级时间自注意力模型。如需了解更多细节，请参阅预印本论文《Land Cover Classification with Gaussian Processes using spatio-spectro-temporal features》的VI-A-1章节，该论文可在此处获取。相关模型的实现代码可在开源代码库中获取。