lvis

  • תיאור :

LVIS: מערך נתונים לפילוח מופעי אוצר מילים גדולים.

  • תיעוד נוסף : חקור על ניירות עם קוד

  • דף הבית : https://www.lvisdataset.org/

  • קוד מקור : tfds.datasets.lvis.Builder

  • גרסאות :

    • 1.1.0 : נוספו שדות neg_category_ids ו- not_exhaustive_category_ids .
    • 1.2.0 : נוספו שמות מחלקות.
    • 1.3.0 : נוסף פיצול מיניבל.
    • 1.4.0 (ברירת מחדל) : נוספו מסכות פילוח לפיצול המיניבל.
  • גודל הורדה : 25.38 GiB

  • גודל מערך נתונים : 23.06 GiB

  • שמירה אוטומטית במטמון ( תיעוד ): לא

  • פיצולים :

לְפַצֵל דוגמאות
'minival' 4,809
'test' 19,822
'train' 100,170
'validation' 19,809
  • מבנה תכונה :
FeaturesDict({
    'image': Image(shape=(None, None, 3), dtype=uint8),
    'image/id': int64,
    'neg_category_ids': Sequence(ClassLabel(shape=(), dtype=int64, num_classes=1203)),
    'not_exhaustive_category_ids': Sequence(ClassLabel(shape=(), dtype=int64, num_classes=1203)),
    'objects': Sequence({
        'area': int64,
        'bbox': BBoxFeature(shape=(4,), dtype=float32),
        'id': int64,
        'label': ClassLabel(shape=(), dtype=int64, num_classes=1203),
        'segmentation': Image(shape=(None, None, 1), dtype=uint8),
    }),
})
  • תיעוד תכונה :
תכונה מַחלָקָה צוּרָה Dtype תֵאוּר
FeaturesDict
תְמוּנָה תְמוּנָה (אין, אין, 3) uint8
תמונה/מזהה מוֹתֵחַ int64
neg_category_ids Sequence (ClassLabel) (אַף לֹא אֶחָד,) int64
מזהי_קטגוריה_לא_מצמצים Sequence (ClassLabel) (אַף לֹא אֶחָד,) int64
חפצים רֶצֶף
חפצים/שטח מוֹתֵחַ int64
אובייקטים/bbox BBoxFeature (4,) לצוף32
אובייקטים/מזהה מוֹתֵחַ int64
חפצים/תווית ClassLabel int64
אובייקטים/פילוח תְמוּנָה (אין, אין, 1) uint8

רְאִיָה

  • ציטוט :
@inproceedings{gupta2019lvis,
  title={ {LVIS}: A Dataset for Large Vocabulary Instance Segmentation},
  author={Gupta, Agrim and Dollar, Piotr and Girshick, Ross},
  booktitle={Proceedings of the {IEEE} Conference on Computer Vision and Pattern Recognition},
  year={2019}
}