resaro/eurosat

EuroSAT Image Classification Dataset

This dataset contains the EuroSAT satellite image classification data in parquet format for easy loading and processing.

Dataset Information

• —Task: Image Classification
• —Source: EuroSAT Dataset
• —Classes: 10 land use/land cover classes
• —Image Size: 64x64 pixels (RGB)
• —Format: Parquet with embedded images
• —Splits: train, test

Classes

The dataset contains 10 land use and land cover classes:

Usage

from datasets import load_dataset

# Load the dataset
ds = load_dataset("resaro/eurosat")

# Access splits
print(ds["train"][0])  # First training example
print(ds["test"][0])   # First test example

# Iterate over the dataset
for example in ds["train"]:
    image = example["image"]  # PIL Image
    label = example["label"]  # Integer 0-9
    # Your processing here

Dataset Structure

Each example contains:

• —image: PIL Image object (64x64 RGB)
• —label: Integer label (0-9) corresponding to the class

Data Splits

Class Distribution (Training Set)

All classes are balanced with approximately 99 samples per class in the training set.

Citation

If you use this dataset, please cite the original EuroSAT paper:

@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},
  volume={12},
  number={7},
  pages={2217--2226},
  year={2019},
  publisher={IEEE}
}

License

MIT License - Please refer to the original EuroSAT dataset for detailed license information.