intelligent

  • Description :

CLEVR est un ensemble de données de diagnostic qui teste une gamme de capacités de raisonnement visuel. Il contient un minimum de biais et comporte des annotations détaillées décrivant le type de raisonnement requis par chaque question.

Diviser Exemples
'test' 15 000
'train' 70 000
'validation' 15 000
  • Structure des fonctionnalités :
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),
    }),
})
  • Documentation des fonctionnalités :
Fonctionnalité Classe Forme Type D Description
FonctionnalitésDict
nom de fichier Texte chaîne
image Image (Aucun, Aucun, 3) uint8
objets Séquence
objets/3d_coords Tenseur (3,) flotteur32
objets/couleur Étiquette de classe int64
objets/matériel Étiquette de classe int64
objets/pixel_coords Tenseur (3,) flotteur32
objets/rotation Tenseur flotteur32
objets/forme Étiquette de classe int64
objets/taille Étiquette de classe int64
Question Réponse Séquence
question_réponse/réponse Texte chaîne
question_réponse/question Texte chaîne

Visualisation

  • Citation :
@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}
}