EuroSAT

Task: Land Use Classification Classes: 10 Image Size: 64x64x3

Overview

EuroSAT is a land use / land cover classification dataset built from Sentinel-2 satellite imagery, covering 10 classes with 27,000 labeled 64x64 RGB images in total.

  • Train: all 27,000 images (no official split ships with the archive)

There is no official train/val/test split, so this builder exposes the full dataset as a single train split (the same handling used for AWA2). Hold out your own validation set if you need one.

../../_images/eurosat_teaser.png

Data Structure

When accessing an example using ds[i], you will receive a dictionary with the following keys:

Key

Type

Description

image

PIL.Image.Image

RGB image, 64x64

label

int

Land use / land cover class (0-9)

Classes

AnnualCrop, Forest, HerbaceousVegetation, Highway, Industrial, Pasture, PermanentCrop, Residential, River, SeaLake

Usage Example

from stable_datasets.images.eurosat import EuroSAT

ds = EuroSAT(split="train")

sample = ds[0]
print(sample.keys())  # {"image", "label"}

class_names = ds.info.features["label"].names
print(class_names[sample["label"]])

# Optional: make it PyTorch-friendly
ds_torch = ds.with_format("torch")

References

  • Official website: https://github.com/phelber/EuroSAT

  • License: MIT License (Sentinel-2 imagery is also subject to the Copernicus Sentinel Data Terms and Conditions)

  • Paper: EuroSAT: A Novel Dataset and Deep Learning Benchmark for Land Use and Land Cover Classification (IEEE JSTARS 2019)

Citation

@article{helber2019eurosat,
  title={EuroSAT: A Novel Dataset and Deep Learning Benchmark for Land Use and Land Cover Classification},
  author={Helber, Patrick and Bischke, Benjamin and Dengel, Andreas and Borth, Damian},
  journal={IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing},
  year={2019}
}