ויזואליזציה: חקור ב מכיר את נתוני שלך
תיאור:
ImageNet-A היא קבוצת תמונות המסומנות עם תוויות ImageNet שהושגו על ידי איסוף נתונים חדשים ושמירה רק על התמונות שדגמי ResNet-50 אינם מסוגלים לסווג כראוי. לפרטים נוספים עיינו בעיתון.
שטח התווית זהה לזה של ImageNet2012. כל דוגמה מיוצגת כמילון עם המקשים הבאים:
- 'תמונה': התמונה, חותך (H, W, 3).
- 'label': מספר שלם בטווח [0, 1000).
'file_name': עוקץ ייחודי המזהה את הדוגמה בתוך מערך הנתונים.
קוד מקור:
tfds.image_classification.ImagenetA
גרסאות:
-
0.1.0
(ברירת המחדל): הערות השחרור לא.
-
גודל ההורדה:
655.70 MiB
מערך נתונים גודל:
650.87 MiB
Auto-במטמון ( תיעוד ): אין
פיצולים:
לְפַצֵל | דוגמאות |
---|---|
'test' | 7,500 |
- מאפיינים:
FeaturesDict({
'file_name': Text(shape=(), dtype=tf.string),
'image': Image(shape=(None, None, 3), dtype=tf.uint8),
'label': ClassLabel(shape=(), dtype=tf.int64, num_classes=1000),
})
מפתחות השגחה (ראה
as_supervised
doc ):('image', 'label')
איור ( tfds.show_examples ):
- דוגמאות ( tfds.as_dataframe ):
- ציטוט:
@article{hendrycks2019nae,
title={Natural Adversarial Examples},
author={Dan Hendrycks and Kevin Zhao and Steven Basart and Jacob Steinhardt and Dawn Song},
journal={arXiv preprint arXiv:1907.07174},
year={2019}
}