open_images_v4

Open Images は、画像レベルのラベルとオブジェクト境界ボックスで注釈が付けられた ~9M の画像のデータセットです。

V4 のトレーニング セットには、174 万個の画像の 600 個のオブジェクト クラスに対応する 1460 万個の境界ボックスが含まれており、オブジェクトの位置の注釈を含む既存のデータセットの中で最大のものになります。ボックスは、正確さと一貫性を確保するために、プロのアノテーターによって大部分が手作業で描かれています。画像は非常に多様で、多くの場合、複数のオブジェクトを含む複雑なシーンが含まれています (画像あたり平均 8.4)。さらに、データセットには、何千ものクラスにまたがる画像レベルのラベルで注釈が付けられています。

スプリット
'test' 125,436
'train' 1,743,042
'validation' 41,620
  • 機能構造:
FeaturesDict({
    'bobjects': Sequence({
        'bbox': BBoxFeature(shape=(4,), dtype=float32),
        'is_depiction': int8,
        'is_group_of': int8,
        'is_inside': int8,
        'is_occluded': int8,
        'is_truncated': int8,
        'label': ClassLabel(shape=(), dtype=int64, num_classes=601),
        'source': ClassLabel(shape=(), dtype=int64, num_classes=6),
    }),
    'image': Image(shape=(None, None, 3), dtype=uint8),
    'image/filename': Text(shape=(), dtype=string),
    'objects': Sequence({
        'confidence': int32,
        'label': ClassLabel(shape=(), dtype=int64, num_classes=19995),
        'source': ClassLabel(shape=(), dtype=int64, num_classes=6),
    }),
    'objects_trainable': Sequence({
        'confidence': int32,
        'label': ClassLabel(shape=(), dtype=int64, num_classes=7186),
        'source': ClassLabel(shape=(), dtype=int64, num_classes=6),
    }),
})
  • 機能のドキュメント:
特徴クラスDtype説明
特徴辞書
物体順序
bobjects/bbox BBoxFeature (4) float32
bobjects/is_depictionテンソルint8
bobjects/is_group_ofテンソルint8
bobjects/is_insideテンソルint8
bobjects/is_occludedテンソルint8
bobjects/is_truncatedテンソルint8
オブジェクト/ラベルクラスラベルint64
オブジェクト/ソースクラスラベルint64
画像画像(なし、なし、3) uint8
画像/ファイル名文章ストリング
オブジェクト順序
オブジェクト/自信テンソルint32
オブジェクト/ラベルクラスラベルint64
オブジェクト/ソースクラスラベルint64
objects_trainable順序
objects_trainable/信頼テンソルint32
objects_trainable/ラベルクラスラベルint64
objects_trainable/ソースクラスラベルint64
@article{OpenImages,
  author = {Alina Kuznetsova and
            Hassan Rom and
            Neil Alldrin and
            Jasper Uijlings and
            Ivan Krasin and
            Jordi Pont-Tuset and
            Shahab Kamali and
            Stefan Popov and
            Matteo Malloci and
            Tom Duerig and
            Vittorio Ferrari},
  title = {The Open Images Dataset V4: Unified image classification,
           object detection, and visual relationship detection at scale},
  year = {2018},
  journal = {arXiv:1811.00982}
}
@article{OpenImages2,
  author = {Krasin, Ivan and
            Duerig, Tom and
            Alldrin, Neil and
            Ferrari, Vittorio
            and Abu-El-Haija, Sami and
            Kuznetsova, Alina and
            Rom, Hassan and
            Uijlings, Jasper and
            Popov, Stefan and
            Kamali, Shahab and
            Malloci, Matteo and
            Pont-Tuset, Jordi and
            Veit, Andreas and
            Belongie, Serge and
            Gomes, Victor and
            Gupta, Abhinav and
            Sun, Chen and
            Chechik, Gal and
            Cai, David and
            Feng, Zheyun and
            Narayanan, Dhyanesh and
            Murphy, Kevin},
  title = {OpenImages: A public dataset for large-scale multi-label and
           multi-class image classification.},
  journal = {Dataset available from
             https://storage.googleapis.com/openimages/web/index.html},
  year={2017}
}

open_images_v4/original (デフォルト設定)

  • 構成の説明: 元の解像度と品質の画像。

  • データセットサイズ: 562.42 GiB

  • ( tfds.show_examples ):

視覚化

open_images_v4/300k

  • 構成の説明: 画像は約 300,000 ピクセルで、72 JPEG 品質です。

  • データセットサイズ: 81.92 GiB

  • ( tfds.show_examples ):

視覚化

open_images_v4/200k

  • 構成の説明: 画像は約 200,000 ピクセルで、72 JPEG 品質です。

  • データセットサイズ: 60.70 GiB

  • ( tfds.show_examples ):

視覚化