SEN12MS-CR: Sentinel-1/Sentinel-2 Cloud Removal Dataset (Parquet mirror)

Provider
Ebel, Meraner, Schmitt, Zhu (Technical University of Munich); Parquet mirror by Hermanni
License
CC-BY-4.0 (source)
Size
177 files
Format
Parquet (in sampled files)
Stored at
Hugging Face (https://huggingface.co/datasets/Hermanni/sen12mscr)
Access
Free and public. No account with the data provider is needed.
DOI
10.1109/TGRS.2020.3024744
Category
Earth Observation & Geospatial
Source
https://huggingface.co/datasets/Hermanni/sen12mscr

About SEN12MS-CR: Sentinel-1/Sentinel-2 Cloud Removal Dataset (Parquet mirror)

SEN12MS-CR is the standard benchmark for cloud removal from optical satellite imagery using co-registered SAR. Each sample is a triplet over the same 256x256 pixel footprint at 10 m: a Sentinel-1 dual-polarisation (VV, VH) SAR patch, a cloudy Sentinel-2 Level-1C patch with all 13 bands, and a cloud-free Sentinel-2 target from a nearby date. The patches come from 169 regions of interest sampled worldwide across all four seasons, following the SEN12MS design, and total about 122,000 triplets in the original release by Ebel et al. (IEEE TGRS, 2021). This repository is a reorganised mirror of the TUM mediaTUM release in Parquet, split into spring/, summer/, fall/ and winter/ folders; SAR is float32, optical arrays int16 with shape stored per row (HWC 256x256x13), and it loads with datasets.load_dataset in streaming mode. About 390 GB. The original is published by TUM under CC BY 4.0; cite Ebel et al. 2021.

How to load this dataset in Python

from datasets import load_dataset
ds = load_dataset("Hermanni/sen12mscr")
print(ds)

Data structure

File types in sampled files: Parquet

Files

Example file: fall/scene_1.parquet

Analyze this dataset without downloading it

On Scigantic, SEN12MS-CR: Sentinel-1/Sentinel-2 Cloud Removal Dataset (Parquet mirror) opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.

This page is also available as Markdown and as JSON.

Related Earth Observation & Geospatial datasets