- Descrição :
CLEVR é um conjunto de dados de diagnóstico que testa uma variedade de habilidades de raciocínio visual. Ele contém preconceitos mínimos e anotações detalhadas que descrevem o tipo de raciocínio que cada pergunta exige.
Documentação adicional : Explore artigos com código
Página inicial : https://cs.stanford.edu/people/jcjohns/clevr/
Código fonte :
tfds.datasets.clevr.BuilderVersões :
-
3.0.0: Sem notas de lançamento. -
3.1.0(padrão): Adicione texto de pergunta/resposta.
-
Tamanho do download :
17.72 GiBTamanho do conjunto de dados :
17.75 GiBArmazenado em cache automaticamente ( documentação ): Não
Divisões :
| Dividir | Exemplos |
|---|---|
'test' | 15.000 |
'train' | 70.000 |
'validation' | 15.000 |
- Estrutura de recursos :
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),
}),
})
- Documentação de recursos :
| Recurso | Aula | Forma | Tipo D | Descrição |
|---|---|---|---|---|
| RecursosDict | ||||
| nome do arquivo | Texto | corda | ||
| imagem | Imagem | (Nenhum, Nenhum, 3) | uint8 | |
| objetos | Seqüência | |||
| objetos/3d_coords | Tensor | (3,) | float32 | |
| objetos/cor | ClassLabel | int64 | ||
| objetos/materiais | ClassLabel | int64 | ||
| objetos/coords_pixel | Tensor | (3,) | float32 | |
| objetos/rotação | Tensor | float32 | ||
| objetos/forma | ClassLabel | int64 | ||
| objetos/tamanho | ClassLabel | int64 | ||
| Resposta da questão | Seqüência | |||
| pergunta_resposta/resposta | Texto | corda | ||
| pergunta_resposta/pergunta | Texto | corda |
Chaves supervisionadas (consulte o documento
as_supervised):NoneFigura ( tfds.show_examples ):

- Exemplos ( tfds.as_dataframe ):
- Citação :
@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}
}