coco_captions

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。このバージョンには、COCO 2014 の画像、境界ボックス、ラベル、キャプションが含まれており、Karpathy と Li (2015) によって定義されたサブセットに分割されています。これにより、元の COCO 2014 検証データが、新しい 5000 枚の画像の検証およびテスト セットと、残りの約 3 万枚の画像を含む「restval」セットに効果的に分割されます。すべての分割にはキャプション アノテーションがあります。

スプリット
'restval' 30,504
'test' 5,000
'train' 82,783
'val' 5,000
  • 機能構造:
FeaturesDict({
    'captions': Sequence({
        'id': int64,
        'text': string,
    }),
    '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
キャプション順序
キャプション/idテンソルint64
キャプション/テキストテンソルストリング
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

  • 引用
@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}
}@inproceedings{DBLP:conf/cvpr/KarpathyL15,
  author    = {Andrej Karpathy and
               Fei{-}Fei Li},
  title     = {Deep visual-semantic alignments for generating image
               descriptions},
  booktitle = { {IEEE} Conference on Computer Vision and Pattern Recognition,
               {CVPR} 2015, Boston, MA, USA, June 7-12, 2015},
  pages     = {3128--3137},
  publisher = { {IEEE} Computer Society},
  year      = {2015},
  url       = {https://doi.org/10.1109/CVPR.2015.7298932},
  doi       = {10.1109/CVPR.2015.7298932},
  timestamp = {Wed, 16 Oct 2019 14:14:50 +0200},
  biburl    = {https://dblp.org/rec/conf/cvpr/KarpathyL15.bib},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}

coco_captions/2014 (デフォルト設定)