User: Guest  Login
Title:

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

Document type:
Forschungsdaten
Publication date:
14.06.2022
Responsible:
Zhu, Xiaoxiang
Authors:
Wang, Yi; Ait Ali Braham, Nassim; Albrecht, Conrad M; Xiong, Zhitong; Liu, Chenying; Zhu, Xiaoxiang
Author affiliation:
TUM
Publisher:
TUM
Identifier:
doi:10.14459/2022mp1660427.001
Concept DOI:
doi:10.14459/2022mp1660427
End date of data production:
31.03.2022
Subject area:
DAT Datenverarbeitung, Informatik; GEO Geowissenschaften
Resource type:
Abbildungen von Objekten / image of objects
Data type:
Bilder / images
Description:
The SSL4EO-S12 dataset is a large-scale dataset for unsupervised/self-supervised 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 covering 2640mx2640m and including four seasonal time stamps. The raw dataset is provided in GeoTiff format, with each band being one single file.
Method of data assessment:
Automatic download, image preparation and processing using Google Earth Engine and Python
Links:

Additional information: https://github.com/zhu-xlab/SSL4EO-S12

A compressed version (Sentinel-2 8-bit, Sentinel-1 8-bit) of the data can be found under: https://mediatum.ub.tum.de/1702379
Key words:
Self-supervised Learning; Remote Sensing; Earth Observation; Sentinel-1/2
Technical remarks:
View and download (1,5 TB total, 3 Files)
The data server also offers downloads with FTP
The data server also offers downloads with rsync (password m1660427.001):
rsync rsync://m1660427.001@dataserv.ub.tum.de/m1660427.001/
Language:
en
Rights:
by, http://creativecommons.org/licenses/by/4.0
 BibTeX