MajorTOM-Core

MajorTOM-Core是由欧洲空间局Φ-lab创建的一个大规模、开放访问的地表观测数据集，旨在为地球观测（EO）模型提供丰富的训练资源。该数据集覆盖了地球陆地表面的绝大部分区域，包含超过2.5万亿像素的Sentinel-2影像数据，数据大小和覆盖范围均创历史新高。创建过程中，采用了先进的地理索引系统和质量控制措施，确保数据的高质量和可用性。MajorTOM-Core不仅是一个数据集，更是一个框架，为未来更多数据集的整合和扩展提供了标准和基础。该数据集适用于深度学习模型的训练，特别是在需要大量地理空间数据的场景中，如地理感知学习、土地利用监测等。

MajorTOM-Core is a large-scale, openly accessible land surface observation dataset developed by European Space Agency's Φ-lab, designed to provide rich training resources for Earth Observation (EO) models. This dataset covers most terrestrial areas across the globe, contains over 2.5 trillion pixels of Sentinel-2 imagery, and both its data volume and coverage have set new historical records. During its development, advanced geospatial indexing systems and quality control measures were employed to guarantee the high quality and usability of the dataset. Beyond being a mere dataset, MajorTOM-Core also serves as a framework that establishes standards and foundations for the integration and expansion of future datasets. It is applicable for training deep learning models, particularly in scenarios demanding large volumes of geospatial data, such as geography-aware learning, land use monitoring, and other related applications.