Albumentations integration in PyTorch - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to apply Albumentations transformations to an image tensor.

PyTorch

import albumentations as A
from albumentations.pytorch import ToTensorV2

transform = A.Compose([
    A.HorizontalFlip(p=[1]),
    ToTensorV2()
])

Drag options to blanks, or click blank then click option'

A2

B0.5

C-0.5

D1.5

Attempts:

3 left

💡 Hint

Common Mistakes

Using a probability greater than 1 causes an error.

Using a negative probability is invalid.

The probability parameter p in Albumentations must be between 0 and 1, so 0.5 is valid.

2fill in blank

medium

Complete the code to convert a PIL image to a numpy array before applying Albumentations.

PyTorch

from PIL import Image
import numpy as np

image = Image.open('image.jpg')
image_np = [1]

Drag options to blanks, or click blank then click option'

Aimage.numpy()

Bimage.toarray()

Cnp.asarray(image.convert())

Dnp.array(image)

Attempts:

3 left

💡 Hint

Common Mistakes

Calling non-existent methods like toarray() or numpy() on PIL images.

Using np.asarray with incomplete arguments.

To convert a PIL Image to a numpy array, use np.array(image).

3fill in blank

hard

Fix the error in applying Albumentations transform to a batch of images.

PyTorch

batch_transformed = [transform(image=[1]) for image in batch_images]

Drag options to blanks, or click blank then click option'

Anp.array(image)

Bimage

Cimage.numpy()

Dimage.to_tensor()

Attempts:

3 left

💡 Hint

Common Mistakes

Passing PIL images directly without conversion.

Using non-existent methods like to_tensor() or numpy() on images.

Albumentations expects images as numpy arrays, so convert each image with np.array(image).

4fill in blank

hard

Fill both blanks to create a custom PyTorch Dataset that applies Albumentations transforms.

PyTorch

from torch.utils.data import Dataset

class CustomDataset(Dataset):
    def __init__(self, images, labels, transform=None):
        self.images = images
        self.labels = labels
        self.transform = [1]

    def __getitem__(self, idx):
        image = self.images[idx]
        label = self.labels[idx]
        if self.transform:
            image = self.transform(image=[2])['image']
        return image, label

Drag options to blanks, or click blank then click option'

Atransform

Bimages

Cnp.array(image)

Dlabels

Attempts:

3 left

💡 Hint

Common Mistakes

Not storing the transform in __init__.

Passing the wrong variable to the transform.

The transform is stored in self.transform and applied to the numpy array of the image.

5fill in blank

hard

Fill all three blanks to correctly integrate Albumentations with PyTorch DataLoader.

PyTorch

import torch
from torch.utils.data import DataLoader

transform = A.Compose([
    A.RandomBrightnessContrast(p=[1]),
    ToTensorV2()
])

dataset = CustomDataset(images, labels, transform=[2])
dataloader = DataLoader(dataset, batch_size=[3], shuffle=True)

Drag options to blanks, or click blank then click option'

A0.3

Btransform

C32

Dimages

Attempts:

3 left

💡 Hint

Common Mistakes

Using invalid probability values.

Passing wrong variables to dataset or DataLoader.

Set probability p=0.3, pass the transform object, and use batch size 32 for DataLoader.