EuroSAT

EuroSAT

MH Specific

Sentinel-2 satellite image dataset for land use and land cover classification. Contains geo-referenced RGB and multispectral images across 10 classes. Applicable to Maharashtra remote sensing and agricultural monitoring

Fine-tune a land use classifier pre-trained on EuroSAT for Maharashtra-specific land cover classification.
HomepageHuggingFaceGitHub

Quick Start

from datasets import load_dataset
ds = load_dataset('blanchon/EuroSAT_RGB')
print(f'Samples: {len(ds["train"])}')
from collections import Counter
print(Counter(ds['train']['label']).most_common())
Modality
Satellite imagery (RGB + multispectral)
Size
27K images (64x64 px), 10 classes
License
Format
PNG/JPEG
Language
mr
Update Frequency
static
Organization
German Research Center for Artificial Intelligence (DFKI)
Category
vision-ocr-multimodal

Schema

FieldTypeDescription
imageimageSentinel-2 satellite image patch (64x64 pixels)
labelstringLand use class (10 classes)

Build With This

Create a transfer learning pipeline adapting European-trained satellite classifiers to Indian/Maharashtra contexts
Develop a few-shot land cover classifier using EuroSAT as pre-training for limited Maharashtra training samples
Build a satellite image labeling tool that uses EuroSAT-trained models as initial predictions for Maharashtra data annotation

AI Use Cases

Land cover classificationremote sensingagricultural monitoring
Last verified: 2026-03-07