Random erasing in Computer Vision - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to import the RandomErasing transform from torchvision.

Computer Vision

from torchvision.transforms import [1]

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

ANormalize

BRandomErasing

CResize

DToTensor

Attempts:

3 left

💡 Hint

Common Mistakes

Importing unrelated transforms like Resize or Normalize.

Forgetting to import RandomErasing before using it.

The RandomErasing transform is imported from torchvision.transforms to apply random erasing augmentation.

2fill in blank

medium

Complete the code to create a RandomErasing transform with a probability of 0.5.

Computer Vision

random_erase = RandomErasing(p=[1])

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

A0.5

B0.0

C1.0

D2.0

Attempts:

3 left

💡 Hint

Common Mistakes

Using a probability greater than 1 or less than 0.

Setting probability to 0 means no erasing happens.

The probability p controls how often random erasing is applied. Setting p=0.5 means it applies half the time.

3fill in blank

hard

Fix the error in applying RandomErasing to a tensor image named img.

Computer Vision

erased_img = random_erase([1])

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

Aimg.numpy()

Bimg.shape

Cimg

Dimg.tolist()

Attempts:

3 left

💡 Hint

Common Mistakes

Passing numpy arrays or lists instead of tensor images.

Passing the shape or other attributes instead of the image tensor.

RandomErasing expects a tensor image input. Passing img directly works if it is a tensor.

4fill in blank

hard

Fill both blanks to create a RandomErasing transform with scale between 0.02 and 0.33.

Computer Vision

random_erase = RandomErasing(p=0.5, scale=([1], [2]))

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

A0.02

B0.5

C0.33

D1.0

Attempts:

3 left

💡 Hint

Common Mistakes

Swapping the scale values or using values outside 0-1.

Using the probability value instead of scale values.

The scale parameter defines the range of erased area proportion. Here, it is set from 0.02 to 0.33.

5fill in blank

hard

Fill all three blanks to create a RandomErasing transform with probability 0.7, ratio between 0.3 and 3.3, and value 0.

Computer Vision

random_erase = RandomErasing(p=[1], ratio=([2], [3]), value=0)

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

A0.5

B0.7

C0.3

D3.3

Attempts:

3 left

💡 Hint

Common Mistakes

Using incorrect probability values outside 0-1.

Swapping ratio values or using invalid ranges.

The p parameter sets the probability to 0.7, ratio sets the aspect ratio range from 0.3 to 3.3, and value=0 means erased pixels are set to zero.