CIFAR100 small images classification dataset

load_data function

keras.datasets.cifar100.load_data(label_mode="fine")

Loads the CIFAR100 dataset.

This is a dataset of 50,000 32×32 color training images and 10,000 test images, labeled over 100 fine-grained classes that are grouped into 20 coarse-grained classes. See more info at the CIFAR homepage.

Arguments

• label_mode: one of "fine", "coarse". If it is "fine", the category labels are the fine-grained labels, and if it is "coarse", the output labels are the coarse-grained superclasses.

Returns

• Tuple of NumPy arrays: (x_train, y_train), (x_test, y_test).

x_train: uint8 NumPy array of grayscale image data with shapes (50000, 32, 32, 3), containing the training data. Pixel values range from 0 to 255.

y_train: uint8 NumPy array of labels (integers in range 0-99) with shape (50000, 1) for the training data.

x_test: uint8 NumPy array of grayscale image data with shapes (10000, 32, 32, 3), containing the test data. Pixel values range from 0 to 255.

y_test: uint8 NumPy array of labels (integers in range 0-99) with shape (10000, 1) for the test data.

Example

(x_train, y_train), (x_test, y_test) = keras.datasets.cifar100.load_data()
assert x_train.shape == (50000, 32, 32, 3)
assert x_test.shape == (10000, 32, 32, 3)
assert y_train.shape == (50000, 1)
assert y_test.shape == (10000, 1)