joeyism / py-cifar10

This library was created to allow an easy usage of CIFAR 10 DATA. This is a wrapper around the instructions givn on the CIFAR 10 site

CIFAR 10 Dataset Library

This library was created to allow an easy usage of CIFAR 10 DATA. This is a wrapper around the instructions given on the CIFAR 10 Site

Installation

pip3 install cifar10

Sample Usage

import cifar10

# Train data
for image, label in cifar10.data_batch_generator():
    image # numpy array of an image, which is of shape 32 x 32 x 3
    label # integer value of the image label

# Test data
for image, label in cifar10.test_batch_generator():
    image # numpy array of an image, which is of shape 32 x 32 x 3
    label # integer value of the image label

API

data_batch_generator

Returns a generator of each image and label pair for data batch

data_batch_generator(cache_location: str=".") -> Iterator[Tuple[np.array, int]]

parameters

• cache_location (default: library folder location): where to cache the cifar10 data

test_batch_generator

Returns a generator of each image and label pair for test batch

test_batch_generator(cache_location: str=".") -> Iterator[Tuple[np.array, int]]

parameters

• cache_location (default: library folder location): where to cache the cifar10 data

meta

Returns the raw meta file

meta(cache_location: str=".") -> Dict[bytes, Any]

parameters

• cache_location (default: library folder location): where to cache the cifar10 data

image_label_map

Returns a dictionary of label

{
    0: 'airplane',
    1: 'automobile',
    ...
}

image_label_map(cache_location: str=".") -> Dict[int, str]

parameters

• cache_location (default: library folder location): where to cache the cifar10 data