باهوش

CLEVR یک مجموعه داده تشخیصی است که طیف وسیعی از توانایی های استدلال بصری را آزمایش می کند. این شامل حداقل سوگیری است و دارای حاشیه نویسی های دقیق است که نوع استدلال مورد نیاز هر سوال را توصیف می کند.

شکاف مثال ها
'test' 15000
'train' 70000
'validation' 15000
  • ساختار ویژگی :
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),
    }),
})
  • مستندات ویژگی :
ویژگی کلاس شکل نوع D شرح
FeaturesDict
نام فایل متن رشته
تصویر تصویر (هیچ، هیچ، 3) uint8
اشیاء توالی
اشیاء/3d_coords تانسور (3،) float32
اشیاء / رنگ ClassLabel int64
اشیاء / مواد ClassLabel int64
اشیاء/pixel_coords تانسور (3،) float32
اشیاء / چرخش تانسور float32
اشیاء/شکل ClassLabel int64
اشیاء/اندازه ClassLabel int64
سوال_پاسخ توالی
سوال_پاسخ/جواب متن رشته
سوال_پاسخ/سوال متن رشته

تجسم

  • نقل قول :
@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}
}