- Opis :
LVIS: Zbiór danych do segmentacji wystąpień dużego słownika.
Dodatkowa dokumentacja : Przeglądaj dokumenty z kodem
Strona główna : https://www.lvisdataset.org/
Kod źródłowy :
tfds.datasets.lvis.Builder
Wersje :
-
1.1.0
: Dodano polaneg_category_ids
inot_exhaustive_category_ids
. -
1.2.0
: Dodano nazwy klas. -
1.3.0
: Dodano podział minival. -
1.4.0
(domyślne) : Dodano maski segmentacji do podziału minival.
-
Rozmiar pobierania :
25.38 GiB
Rozmiar zbioru danych :
23.06 GiB
Automatyczne buforowanie ( dokumentacja ): Nie
Podziały :
Podział | Przykłady |
---|---|
'minival' | 4809 |
'test' | 19 822 |
'train' | 100 170 |
'validation' | 19 809 |
- Struktura funkcji :
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),
}),
})
- Dokumentacja funkcji :
Funkcja | Klasa | Kształt | Typ danych | Opis |
---|---|---|---|---|
FunkcjeSłownik | ||||
obraz | Obraz | (Brak, Brak, 3) | uint8 | |
obraz/id | Napinacz | int64 | ||
neg_category_ids | Sekwencja (etykieta klasy) | (Nic,) | int64 | |
niepełne_id_kategorii | Sekwencja (etykieta klasy) | (Nic,) | int64 | |
obiekty | Sekwencja | |||
obiekty/obszar | Napinacz | int64 | ||
obiekty/bbox | Funkcja BBox | (4,) | float32 | |
obiekty/id | Napinacz | int64 | ||
obiekty/etykieta | Etykieta klasy | int64 | ||
obiekty/segmentacja | Obraz | (Brak, Brak, 1) | uint8 |
Klucze nadzorowane (zobacz dokument
as_supervised
):None
Rysunek ( tfds.show_examples ):
- Przykłady ( tfds.as_dataframe ):
- Cytat :
@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}
}