كوكو

COCO عبارة عن مجموعة بيانات واسعة النطاق لاكتشاف الكائنات وتقسيمها والتعليق عليها.

@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}
}

كوكو / 2014 (التكوين الافتراضي)

  • وصف التكوين : يحتوي هذا الإصدار على صور ومربعات إحاطة وتسميات لإصدار 2014.

  • حجم التحميل : 37.57 GiB

  • حجم مجموعة البيانات : 37.35 GiB

  • الانقسامات :

انشق، مزق أمثلة
'test' 40775
'test2015' 81.434
'train' 82،783
'validation' 40504
  • هيكل الميزة :
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),
    }),
})
  • وثائق الميزة :
ميزة فصل شكل نوع وصف
الميزات
صورة صورة (لا شيء ، لا شيء ، 3) uint8
الصورة / اسم الملف نص سلسلة
الصورة / معرف موتر int64
أشياء تسلسل
الأشياء / المنطقة موتر int64
كائنات / bbox ميزة BBox (4 ،) تعويم 32
الكائنات / معرف موتر int64
كائنات / is_crowd موتر منطقي
الكائنات / التسمية ClassLabel int64

التصور

كوكو / 2017

  • وصف التكوين : يحتوي هذا الإصدار على صور ومربعات إحاطة وتسميات لإصدار 2017.

  • حجم التحميل : 25.20 GiB

  • حجم مجموعة البيانات : 24.98 GiB

  • الانقسامات :

انشق، مزق أمثلة
'test' 40،670
'train' 118،287
'validation' 5000
  • هيكل الميزة :
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),
    }),
})
  • وثائق الميزة :
ميزة فصل شكل نوع وصف
الميزات
صورة صورة (لا شيء ، لا شيء ، 3) uint8
الصورة / اسم الملف نص سلسلة
الصورة / معرف موتر int64
أشياء تسلسل
الأشياء / المنطقة موتر int64
كائنات / bbox ميزة BBox (4 ،) تعويم 32
الكائنات / معرف موتر int64
كائنات / is_crowd موتر منطقي
الكائنات / التسمية ClassLabel int64

التصور

coco / 2017_panoptic

  • وصف التكوين : يحتوي هذا الإصدار على صور ومربعات إحاطة وتسميات لإصدار 2017.

  • حجم التحميل : 19.57 GiB

  • حجم مجموعة البيانات : 19.63 GiB

  • الانقسامات :

انشق، مزق أمثلة
'train' 118،287
'validation' 5000
  • هيكل الميزة :
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),
    }),
})
  • وثائق الميزة :
ميزة فصل شكل نوع وصف
الميزات
صورة صورة (لا شيء ، لا شيء ، 3) uint8
الصورة / اسم الملف نص سلسلة
الصورة / معرف موتر int64
panoptic_image صورة (لا شيء ، لا شيء ، 3) uint8
panoptic_image / اسم الملف نص سلسلة
panoptic_objects تسلسل
panoptic_objects / المنطقة موتر int64
panoptic_objects / bbox ميزة BBox (4 ،) تعويم 32
panoptic_objects / معرف موتر int64
panoptic_objects / is_crowd موتر منطقي
panoptic_objects / التسمية ClassLabel int64