ココ

COCO は、大規模なオブジェクト検出、セグメンテーション、キャプション データセットです。

@article{DBLP:journals/corr/LinMBHPRDZ14,
  author    = {Tsung{-}Yi Lin and
               Michael Maire and
               Serge J. Belongie and
               Lubomir D. Bourdev and
               Ross B. Girshick and
               James Hays and
               Pietro Perona and
               Deva Ramanan and
               Piotr Doll{'{a} }r and
               C. Lawrence Zitnick},
  title     = {Microsoft {COCO:} Common Objects in Context},
  journal   = {CoRR},
  volume    = {abs/1405.0312},
  year      = {2014},
  url       = {http://arxiv.org/abs/1405.0312},
  archivePrefix = {arXiv},
  eprint    = {1405.0312},
  timestamp = {Mon, 13 Aug 2018 16:48:13 +0200},
  biburl    = {https://dblp.org/rec/bib/journals/corr/LinMBHPRDZ14},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}

coco/2014 (デフォルト構成)

  • 構成の説明: このバージョンには、2014 バージョンの画像、境界ボックス、およびラベルが含まれています。

  • ダウンロードサイズ: 37.57 GiB

  • データセットサイズ: 37.35 GiB

  • スプリット:

スプリット
'test' 40,775
'test2015' 81,434
'train' 82,783
'validation' 40,504
  • 機能構造:
FeaturesDict({
    'image': Image(shape=(None, None, 3), dtype=uint8),
    'image/filename': Text(shape=(), dtype=string),
    'image/id': int64,
    'objects': Sequence({
        'area': int64,
        'bbox': BBoxFeature(shape=(4,), dtype=float32),
        'id': int64,
        'is_crowd': bool,
        'label': ClassLabel(shape=(), dtype=int64, num_classes=80),
    }),
})
  • 機能のドキュメント:
特徴クラスDtype説明
特徴辞書
画像画像(なし、なし、3) uint8
画像/ファイル名文章ストリング
画像/IDテンソルint64
オブジェクト順序
オブジェクト/エリアテンソルint64
オブジェクト/bbox BBoxFeature (4) float32
オブジェクト/IDテンソルint64
オブジェクト/is_crowdテンソルブール
オブジェクト/ラベルクラスラベルint64

視覚化

ココ/2017

  • 構成の説明: このバージョンには、2017 バージョンの画像、境界ボックス、およびラベルが含まれています。

  • ダウンロードサイズ: 25.20 GiB

  • データセットサイズ: 24.98 GiB

  • スプリット:

スプリット
'test' 40,670
'train' 118,287
'validation' 5,000
  • 機能構造:
FeaturesDict({
    'image': Image(shape=(None, None, 3), dtype=uint8),
    'image/filename': Text(shape=(), dtype=string),
    'image/id': int64,
    'objects': Sequence({
        'area': int64,
        'bbox': BBoxFeature(shape=(4,), dtype=float32),
        'id': int64,
        'is_crowd': bool,
        'label': ClassLabel(shape=(), dtype=int64, num_classes=80),
    }),
})
  • 機能のドキュメント:
特徴クラスDtype説明
特徴辞書
画像画像(なし、なし、3) uint8
画像/ファイル名文章ストリング
画像/IDテンソルint64
オブジェクト順序
オブジェクト/エリアテンソルint64
オブジェクト/bbox BBoxFeature (4) float32
オブジェクト/IDテンソルint64
オブジェクト/is_crowdテンソルブール
オブジェクト/ラベルクラスラベルint64

視覚化

ココ/2017_panoptic

  • 構成の説明: このバージョンには、2017 バージョンの画像、境界ボックス、およびラベルが含まれています。

  • ダウンロードサイズ: 19.57 GiB

  • データセットサイズ: 19.63 GiB

  • スプリット:

スプリット
'train' 118,287
'validation' 5,000
  • 機能構造:
FeaturesDict({
    'image': Image(shape=(None, None, 3), dtype=uint8),
    'image/filename': Text(shape=(), dtype=string),
    'image/id': int64,
    'panoptic_image': Image(shape=(None, None, 3), dtype=uint8),
    'panoptic_image/filename': Text(shape=(), dtype=string),
    'panoptic_objects': Sequence({
        'area': int64,
        'bbox': BBoxFeature(shape=(4,), dtype=float32),
        'id': int64,
        'is_crowd': bool,
        'label': ClassLabel(shape=(), dtype=int64, num_classes=133),
    }),
})
  • 機能のドキュメント:
特徴クラスDtype説明
特徴辞書
画像画像(なし、なし、3) uint8
画像/ファイル名文章ストリング
画像/IDテンソルint64
panoptic_image画像(なし、なし、3) uint8
panoptic_image/ファイル名文章ストリング
panoptic_objects順序
panoptic_objects/エリアテンソルint64
panoptic_objects/bbox BBoxFeature (4) float32
panoptic_objects/idテンソルint64
panoptic_objects/is_crowdテンソルブール
panoptic_objects/ラベルクラスラベルint64