ユーロサット

  • 説明:

EuroSAT データセットは、Sentinel-2 衛星画像に基づいており、13 のスペクトル バンドをカバーし、27000 のラベル付けされ地理参照されたサンプルを含む 10 のクラスで構成されています。

2 つのデータセットが提供されています。 - rgb: JPEG 画像としてエンコードされた光 R、G、B 周波数帯域のみが含まれます。 - all: 元の値の範囲 (float32) の 13 バンドすべてが含まれます。

URL: https://github.com/phelber/eurosat

スプリット
'train' 27,000
  • 引用
@misc{helber2017eurosat,
    title={EuroSAT: A Novel Dataset and Deep Learning Benchmark for Land Use and Land Cover Classification},
    author={Patrick Helber and Benjamin Bischke and Andreas Dengel and Damian Borth},
    year={2017},
    eprint={1709.00029},
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}

eurosat/rgb (デフォルト構成)

  • 構成の説明: Sentinel-2 RGB チャネル

  • ダウンロードサイズ: 89.91 MiB

  • データセットサイズ: 89.50 MiB

  • 自動キャッシュ(ドキュメント): はい

  • 機能構造:

FeaturesDict({
    'filename': Text(shape=(), dtype=string),
    'image': Image(shape=(64, 64, 3), dtype=uint8),
    'label': ClassLabel(shape=(), dtype=int64, num_classes=10),
})
  • 機能のドキュメント:
特徴クラスDtype説明
特徴辞書
ファイル名文章ストリング
画像画像(64, 64, 3) uint8
ラベルクラスラベルint64

視覚化

ユーロサット/すべて

  • 構成の説明: 13 Sentinel-2 チャネル

  • ダウンロードサイズ: 1.93 GiB

  • データセットサイズ: 5.36 GiB

  • 自動キャッシュ(ドキュメント): いいえ

  • 機能構造:

FeaturesDict({
    'filename': Text(shape=(), dtype=string),
    'label': ClassLabel(shape=(), dtype=int64, num_classes=10),
    'sentinel2': Tensor(shape=(64, 64, 13), dtype=float32),
})
  • 機能のドキュメント:
特徴クラスDtype説明
特徴辞書
ファイル名文章ストリング
ラベルクラスラベルint64
センチネル2テンソル(64, 64, 13) float32