imagenet_a

ImageNet-A は、新しいデータを収集し、ResNet-50 モデルが正しく分類できない画像のみを保持することによって取得された、ImageNet ラベルでラベル付けされた一連の画像です。詳細については、論文を参照してください。

ラベルスペースはImageNet2012と同じです。各例は、次のキーを持つ辞書として表されます。

スプリット
'test' 7,500
  • 機能構造:
FeaturesDict({
    'file_name': Text(shape=(), dtype=string),
    'image': Image(shape=(None, None, 3), dtype=uint8),
    'label': ClassLabel(shape=(), dtype=int64, num_classes=1000),
})
  • 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(なし、なし、3) uint8
ラベルクラスラベルint64

視覚化

  • 引用
@article{hendrycks2019nae,
  title={Natural Adversarial Examples},
  author={Dan Hendrycks and Kevin Zhao and Steven Basart and Jacob Steinhardt and Dawn Song},
  journal={arXiv preprint arXiv:1907.07174},
  year={2019}
}