سيفار 10

تتكون مجموعة بيانات CIFAR-10 من 60000 صورة ملونة 32 × 32 في 10 فئات ، مع 6000 صورة لكل فصل. يوجد 50000 صورة تدريب و 10000 صورة اختبار.

انشق، مزق أمثلة
'test' 10000
'train' 50000
  • هيكل الميزة :
FeaturesDict({
    'id': Text(shape=(), dtype=string),
    'image': Image(shape=(32, 32, 3), dtype=uint8),
    'label': ClassLabel(shape=(), dtype=int64, num_classes=10),
})
  • وثائق الميزة :
ميزة فصل شكل نوع وصف
الميزات
هوية شخصية نص سلسلة
صورة صورة (32 ، 32 ، 3) uint8
ضع الكلمة المناسبة ClassLabel int64

التصور

  • الاقتباس :
@TECHREPORT{Krizhevsky09learningmultiple,
    author = {Alex Krizhevsky},
    title = {Learning multiple layers of features from tiny images},
    institution = {},
    year = {2009}
}