open_images_v4

Open Images عبارة عن مجموعة بيانات من حوالي 9 مليون صورة تم وضع تعليقات توضيحية عليها باستخدام تسميات على مستوى الصورة ومربعات تحيط بالكائنات.

تحتوي مجموعة التدريب لـ V4 على 14.6 مليون مربعات إحاطة لـ 600 فئة كائن على 1.74 مليون صورة ، مما يجعلها أكبر مجموعة بيانات موجودة مع تعليقات توضيحية لموقع الكائن. تم رسم المربعات يدويًا إلى حد كبير من قبل المعلقين المحترفين لضمان الدقة والاتساق. الصور متنوعة للغاية وغالبًا ما تحتوي على مشاهد معقدة بها عدة كائنات (8.4 لكل صورة في المتوسط). علاوة على ذلك ، يتم شرح مجموعة البيانات بملصقات على مستوى الصورة تغطي آلاف الفئات.

انشق، مزق أمثلة
'test' 125436
'train' 1،743،042
'validation' 41620
  • هيكل الميزة :
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),
    }),
})
  • وثائق الميزة :
ميزة فصل شكل نوع وصف
الميزات
bobjects تسلسل
bobjects / bbox ميزة BBox (4 ،) تعويم 32
bobjects / is_depiction موتر int8
bobjects / is_group_of موتر int8
bobjects / is_inside موتر int8
bobjects / is_occluded موتر int8
bobjects / is_truncated موتر int8
bobjects / التسمية ClassLabel int64
bobjects / المصدر ClassLabel int64
صورة صورة (لا شيء ، لا شيء ، 3) uint8
الصورة / اسم الملف نص سلسلة
أشياء تسلسل
الأشياء / الثقة موتر int32
الكائنات / التسمية ClassLabel int64
الكائنات / المصدر ClassLabel int64
Object_trainable تسلسل
object_trainable / الثقة موتر int32
object_trainable / التسمية ClassLabel int64
object_trainable / المصدر ClassLabel 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 (التكوين الافتراضي)

  • وصف التكوين : الصور بالدقة والجودة الأصلية.

  • حجم مجموعة البيانات : 562.42 GiB

  • الشكل ( tfds.show_examples ):

التصور

open_images_v4 / 300k

  • وصف التكوين : تحتوي الصور على 300000 بكسل تقريبًا ، بجودة 72 JPEG.

  • حجم مجموعة البيانات : 81.92 GiB

  • الشكل ( tfds.show_examples ):

التصور

open_images_v4 / 200k

  • وصف التكوين : تحتوي الصور على ما يقرب من 200000 بكسل ، بجودة 72 JPEG.

  • حجم مجموعة البيانات : 60.70 GiB

  • الشكل ( tfds.show_examples ):

التصور