SSL4EO-S12: A Large-scale Multimodal Multitemporal Dataset for Self-supervised Learning in Earth Observation (8-bit)

The SSL4EO-S12 dataset is a large-scale dataset for unsupervised pre-training in Earth observation. The dataset consists of unlabeled patch triplets (Sentinel-1 dual-pol SAR, Sentinel-2 top-of-atmosphere multispectral, Sentinel-2 surface reflectance multispectral) from 251079 locations across the globe. Each patch covers an area of 2640mx2640m and includes four seasonal time stamps. The compressed dataset is provided in normalized 8-bit GeoTiff format, with each band being one single file. Details see <a href=" https://github.com/zhu-xlab/SSL4EO-S12" target="_blank"> https://github.com/zhu-xlab/SSL4EO-S12</a>SSL4EO-S12数据集是一款面向地球观测领域无监督预训练的大规模数据集。该数据集涵盖全球251079个采样点位的未标注影像块三元组，具体包含三类数据：哨兵一号（Sentinel-1）双极化合成孔径雷达（Synthetic Aperture Radar, SAR）影像、哨兵二号（Sentinel-2）大气顶层多光谱影像，以及哨兵二号（Sentinel-2）地表反射率多光谱影像。每个影像块覆盖2640米×2640米的区域，并包含四个季节的时序影像。该压缩数据集以归一化8位GeoTiff格式存储，每个波段对应一个独立文件。详细信息可参阅 <a href="https://github.com/zhu-xlab/SSL4EO-S12" target="_blank">https://github.com/zhu-xlab/SSL4EO-S12</a>。