README.md

metadata

license: mit
dataset_info:
  features:
    - name: image
      dtype: image
    - name: labels
      dtype:
        class_label:
          names:
            '0': AnnualCrop
            '1': Forest
            '2': HerbaceousVegetation
            '3': Highway
            '4': Industrial
            '5': Pasture
            '6': PermanentCrop
            '7': Residential
            '8': River
            '9': SeaLake
  splits:
    - name: train
      num_bytes: 70666516.8
      num_examples: 21600
    - name: validation
      num_bytes: 8700747.8
      num_examples: 2700
    - name: test
      num_bytes: 8631409.1
      num_examples: 2700
  download_size: 93886613
  dataset_size: 87998673.69999999
task_categories:
  - image-classification

Dataset Card for EuroSat

Table of Contents

• How to Use
• Dataset Description
  • Dataset Summary
• Dataset Structure
  • Data Instances
  • Data Fields
  • Data Splits
• Additional Information
  • Licensing Information

How to Use

• Install datasets:

pip install datasets

• How to use in Python

from datasets import load_dataset
train_data = load_dataset("Honaker/eurosat_dataset", split="train")

Dataset Description

• Homepage: https://zenodo.org/record/7711810#.ZAm3k-zMKEA

Dataset Summary

EuroSat is an image classification dataset with 10 different classes on satellite imagery. There is over 27,000 labeled images.

Dataset Structure

The dataset is structured as follows:

DatasetDict({
    train: Dataset({
        features: ['image', 'labels'],
        num_rows: 21600
    })
    validation: Dataset({
        features: ['image', 'labels'],
        num_rows: 2700
    })
    test: Dataset({
        features: ['image', 'labels'],
        num_rows: 2700
    })
})

Data Instances

An example of the data for one image is:

{
 'image': <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=64x64>,
 'labels': 0
}

With the type of each field being defined as:

{
'image': <PIL.JpegImagePlugin.JpegImageFile>,
'labels': Integer

Data Fields

The dataset has the following fields:

• 'image': Satellite image that is of type <PIL.TiffImagePlugin.TiffImageFile image>
• 'labels': the label of the Satellite image as an integer

Data Splits

	Train	Validation	Test
Images	21600	2700	2700

Additional Information

Licensing Information

EuroSat is licensed under a MIT