omníglota

Conjunto de datos Omniglot para el aprendizaje de una sola vez. Este conjunto de datos contiene 1623 caracteres escritos a mano diferentes de 50 alfabetos diferentes.

Separar Ejemplos
'small1' 2,720
'small2' 3,120
'test' 13,180
'train' 19,280
  • Estructura de características :
FeaturesDict({
    'alphabet': ClassLabel(shape=(), dtype=int64, num_classes=50),
    'alphabet_char_id': int64,
    'image': Image(shape=(105, 105, 3), dtype=uint8),
    'label': ClassLabel(shape=(), dtype=int64, num_classes=1623),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
alfabeto Etiqueta de clase int64
alfabeto_char_id Tensor int64
imagen Imagen (105, 105, 3) uint8
etiqueta Etiqueta de clase int64

Visualización

  • Cita :
@article{lake2015human,
  title={Human-level concept learning through probabilistic program induction},
  author={Lake, Brenden M and Salakhutdinov, Ruslan and Tenenbaum, Joshua B},
  journal={Science},
  volume={350},
  number={6266},
  pages={1332--1338},
  year={2015},
  publisher={American Association for the Advancement of Science}
}