Datasets

synopsis

Global datasets available

npu.vision.datasets.CIFAR10 = <npu.core.Dataset.Dataset object>
CIFAR10

CIFAR10 dataset. Contains training/validation/test subsets of data. Accessed using

CIFAR10.train/val/test properties. Can also get direct subset using slices. CIFAR10.train and CIFAR10.val contain the training set of the dataset with a split of 80% and 20% respectively. CIFAR10.test contains 10000 samples from the non-training set of the dataset.

npu.vision.datasets.CocoDetection = <npu.core.Dataset.Dataset object>
CocoDetection

CocoDetection dataset. Contains training/validation/test subsets of data. Accessed using

CocoDetection.train/val/test properties. Can also get direct subset using slices. CocoDetection.train contains the 118287 samples from the train set of the dataset. CocoDetection.val contains the 5000 samples from the val set of the dataset. CocoDetection.test contains the 40670 samples from the test set of the dataset.

npu.vision.datasets.FashionMNIST = <npu.core.Dataset.Dataset object>
FashionMNIST

FashionMNIST dataset. Contains training/validation/test subsets of data. Accessed using

FashionMNIST.train/val/test properties. Can also get direct subset using slices. FashionMNIST.train and FashionMNIST.val contain the training set of the dataset with a split of 80% and 20% respectively. FashionMNIST.test contains the 10000 samples from the non-training set of the dataset.

npu.vision.datasets.KMNIST = <npu.core.Dataset.Dataset object>
KMNIST

KMNIST dataset. Contains training/validation/test subsets of data. Accessed using

KMNIST.train/val/test properties. Can also get direct subset using slices. KMNIST.train and KMNIST.val contain the training set of the dataset with a split of 80% and 20% respectively. KMNIST.test contains 10000 samples from the non-training set of the dataset.

npu.vision.datasets.MNIST = <npu.core.Dataset.Dataset object>
MNIST

MNIST dataset. Contains training/validation/test subsets of data. Accessed using

MNIST.train/val/test properties. Can also get direct subset using slices. MNIST.train and MNIST.val contain the training set of the dataset with a split of 80% and 20% respectively. MNIST.test contains 10000 samples from the non-training set of the dataset.

npu.vision.datasets.QMNIST = <npu.core.Dataset.Dataset object>
QMNIST

QMNIST dataset. Contains training/validation/test subsets of data. Accessed using

QMNIST.train/val/test properties. Can also get direct subset using slices. QMNIST.train and QMNIST.val contain the training set of the dataset with a split of 80% and 20% respectively. QMNIST.test contains the 60000 samples from the non-training set of the dataset.