Create PyTorch datasets and dataset loaders for a subset of CIFAR10 classes.

Select_CIFAR10_Classes.py

import torchvision
import torchvision.transforms as transforms
from torchvision.datasets import CIFAR10
from torch.utils.data import Dataset, DataLoader
import numpy as np

# Transformations
RC = transforms.RandomCrop(32, padding=4)
RHF = transforms.RandomHorizontalFlip()
RVF = transforms.RandomVerticalFlip()
NRM = transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010))
TT = transforms.ToTensor()
TPIL = transforms.ToPILImage()

# Transforms object for trainset with augmentation
transform_with_aug = transforms.Compose([TPIL, RC, RHF, TT, NRM])
# Transforms object for testset with NO augmentation
transform_no_aug = transforms.Compose([TT, NRM])

# Downloading/Louding CIFAR10 data
# , transform = transform_with_aug)
trainset = CIFAR10(root='./data', train=True, download=True)
# , transform = transform_no_aug)
testset = CIFAR10(root='./data', train=False, download=True)
classDict = {'plane': 0, 'car': 1, 'bird': 2, 'cat': 3, 'deer': 4,
             'dog': 5, 'frog': 6, 'horse': 7, 'ship': 8, 'truck': 9}

# Separating trainset/testset data/label
x_train = trainset.data
x_test = testset.data
y_train = trainset.targets
y_test = testset.targets

# Define a function to separate CIFAR classes by class index


def get_class_i(x, y, i):
    """
    x: trainset.train_data or testset.test_data
    y: trainset.train_labels or testset.test_labels
    i: class label, a number between 0 to 9
    return: x_i
    """
    # Convert to a numpy array
    y = np.array(y)
    # Locate position of labels that equal to i
    pos_i = np.argwhere(y == i)
    # Convert the result into a 1-D list
    pos_i = list(pos_i[:, 0])
    # Collect all data that match the desired label
    x_i = [x[j] for j in pos_i]

    return x_i


class DatasetMaker(Dataset):
    def __init__(self, datasets, transformFunc=transform_no_aug):
        """
        datasets: a list of get_class_i outputs, i.e. a list of list of images for selected classes
        """
        self.datasets = datasets
        self.lengths = [len(d) for d in self.datasets]
        self.transformFunc = transformFunc

    def __getitem__(self, i):
        class_label, index_wrt_class = self.index_of_which_bin(self.lengths, i)
        img = self.datasets[class_label][index_wrt_class]
        img = self.transformFunc(img)
        return img, class_label

    def __len__(self):
        return sum(self.lengths)

    def index_of_which_bin(self, bin_sizes, absolute_index, verbose=False):
        """
        Given the absolute index, returns which bin it falls in and which element of that bin it corresponds to.
        """
        # Which class/bin does i fall into?
        accum = np.add.accumulate(bin_sizes)
        if verbose:
            print("accum =", accum)
        bin_index = len(np.argwhere(accum <= absolute_index))
        if verbose:
            print("class_label =", bin_index)
        # Which element of the fallent class/bin does i correspond to?
        index_wrt_class = absolute_index - np.insert(accum, 0, 0)[bin_index]
        if verbose:
            print("index_wrt_class =", index_wrt_class)

        return bin_index, index_wrt_class

# ================== Usage ================== #


# Let's choose cats (class 3 of CIFAR) and dogs (class 5 of CIFAR) as trainset/testset
cat_dog_trainset = \
    DatasetMaker(
        [get_class_i(x_train, y_train, classDict['cat']),
         get_class_i(x_train, y_train, classDict['dog'])],
        transform_with_aug
    )
cat_dog_testset = \
    DatasetMaker(
        [get_class_i(x_test, y_test, classDict['cat']),
         get_class_i(x_test, y_test, classDict['dog'])],
        transform_no_aug
    )

kwargs = {'num_workers': 2, 'pin_memory': False}

# Create datasetLoaders from trainset and testset
trainsetLoader = DataLoader(
    cat_dog_trainset, batch_size=64, shuffle=True, **kwargs)
testsetLoader = DataLoader(
    cat_dog_testset, batch_size=64, shuffle=False, **kwargs)

To test the performance of a neural net at the beginning stages, you don't need all 10 CIFAR-10 classes; 2 is enough and is a lot faster to train. But not any two, the hard pairs such as [cat, dog], [car, truck], and [deer, horse] due to their similarities. In many cases, one can tell the difference between airplanes and horses by comparing measuring blueness of an image! So, I wrote this code to build trainset/testset and trainsetLoader/testsetLoader for a subset of CIFAR-10 classes. If you find this helpful, please give it a star ⭐.

thanks for this!

thank you!
But I got porblem
AttributeError: 'CIFAR10' object has no attribute 'train_data'
how can i fix it?

Thanks for this! Got a small error using this code. Need to convert the output images from numpy to PIL

@hyeongminoh
You could just change those 4 lines as follows.
x_train = trainset.data
x_test = testset.data
y_train = trainset.targets
y_test = testset.targets

• img = Image.fromarray(img) this line is needed before the transformation as @agoel10 said.
• Running your code, I've found out that target values are actually returned wrt ordering of your queries, such as cat and dog above meaning that cat is re-assigned to a target value 0 and dog to 1. How can I keep and return the original target values?

@Seohyeong how do you print trainsetLoader/ testsetLoader target values. Thank you.

I just updated this to work with the latest PyTorch ('1.9.1') and torch vision ('0.10.1').

I did it like this with e for the classes and k for the number of elements, 100 planes, cars and frogs

`dataset = CIFAR10(os.getcwd(), train=True, download=True, transform=transforms.ToTensor())
index = random.choices([i for i ,e in enumerate(list(dataset.targets)) if e == 0], k=100)
index = np.append(index, random.choices([i for i ,e in enumerate(list(dataset.targets)) if e == 1], k=100))
index = np.append(index, random.choices([i for i ,e in enumerate(list(dataset.targets)) if e == 6], k=100))

data = torch.utils.data.Subset(dataset, index)
train_data,val_data, test_data = torch.utils.data.random_split(data,[200,50,50])`