كليفر

CLEVR هي مجموعة بيانات تشخيصية تختبر مجموعة من قدرات التفكير البصري. يحتوي على حد أدنى من التحيزات ويحتوي على شروح تفصيلية تصف نوع التفكير الذي يتطلبه كل سؤال.

انشق، مزق أمثلة
'test' 15000
'train' 70000
'validation' 15000
  • هيكل الميزة :
FeaturesDict({
    'file_name': Text(shape=(), dtype=string),
    'image': Image(shape=(None, None, 3), dtype=uint8),
    'objects': Sequence({
        '3d_coords': Tensor(shape=(3,), dtype=float32),
        'color': ClassLabel(shape=(), dtype=int64, num_classes=8),
        'material': ClassLabel(shape=(), dtype=int64, num_classes=2),
        'pixel_coords': Tensor(shape=(3,), dtype=float32),
        'rotation': float32,
        'shape': ClassLabel(shape=(), dtype=int64, num_classes=3),
        'size': ClassLabel(shape=(), dtype=int64, num_classes=2),
    }),
    'question_answer': Sequence({
        'answer': Text(shape=(), dtype=string),
        'question': Text(shape=(), dtype=string),
    }),
})
  • وثائق الميزة :
ميزة فصل شكل نوع وصف
الميزات
اسم الملف نص سلسلة
صورة صورة (لا شيء ، لا شيء ، 3) uint8
أشياء تسلسل
كائنات / 3d_coords موتر (3 ،) تعويم 32
الكائنات / اللون ClassLabel int64
الأشياء / المواد ClassLabel int64
كائنات / بكسل_كوردس موتر (3 ،) تعويم 32
الأشياء / التناوب موتر تعويم 32
الأشياء / الشكل ClassLabel int64
الأشياء / الحجم ClassLabel int64
جواب السؤال تسلسل
question_answer / الجواب نص سلسلة
question_answer / سؤال نص سلسلة

التصور

  • الاقتباس :
@inproceedings{johnson2017clevr,
  title={ {CLEVR}: A diagnostic dataset for compositional language and elementary visual reasoning},
  author={Johnson, Justin and Hariharan, Bharath and van der Maaten, Laurens and Fei-Fei, Li and Lawrence Zitnick, C and Girshick, Ross},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  year={2017}
}