Image datasets (CIFAR-10, ImageNet) in Computer Vision - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to load the CIFAR-10 dataset using TensorFlow.

Computer Vision

import tensorflow as tf
(cifar_train, cifar_test) = tf.keras.datasets.cifar10.[1]()

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

Aload_data

Bfetch

Cdownload

Dget_dataset

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'download' or 'fetch' which are not valid TensorFlow dataset methods.

Trying to call a method that does not exist in tf.keras.datasets.cifar10.

The correct function to load CIFAR-10 dataset in TensorFlow is load_data(). It returns training and test sets.

2fill in blank

medium

Complete the code to normalize CIFAR-10 images to the range 0-1.

Computer Vision

cifar_train_images = cifar_train[0] / [1]

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

A255

B100

C1

D256

Attempts:

3 left

💡 Hint

Common Mistakes

Dividing by 100 or 256 which does not correctly normalize pixel values.

Not dividing at all, leaving pixel values in 0-255 range.

Image pixel values range from 0 to 255. Dividing by 255 scales them to 0-1.

3fill in blank

hard

Fix the error in this code to load ImageNet dataset using TensorFlow Datasets.

Computer Vision

import tensorflow_datasets as tfds
imagenet_data = tfds.load('imagenet_v2', split=[1])

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

A'train[:80%]'

B'validation'

C'train'

D'test'

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'train' or 'test' which may not exist or are not the validation split.

Using slicing syntax like 'train[:80%]' which is invalid here.

The ImageNet dataset split for validation is named 'validation'. Using 'validation' loads the validation set correctly.

4fill in blank

hard

Fill both blanks to create a dictionary of image shapes and labels from CIFAR-10 training data.

Computer Vision

image_info = {i: {'shape': cifar_train[0][i].[1], 'label': cifar_train[1][i][[2]]} for i in range(5)}

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

Ashape

B0

C1

Dsize

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'size' instead of 'shape' for image dimensions.

Using label index 1 which is out of range.

Each image has a shape attribute for its dimensions. Labels are stored as arrays, so index 0 accesses the label value.

5fill in blank

hard

Fill all three blanks to filter CIFAR-10 images with width greater than 30 and create a dictionary of their labels and shapes.

Computer Vision

filtered = {i: {'label': cifar_train[1][i][0], 'shape': cifar_train[0][i].[1] for i in range(len(cifar_train[0])) if cifar_train[0][i].shape[[2]] [3] 30}

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

Ashape

B1

C>

D0

Attempts:

3 left

💡 Hint

Common Mistakes

Using index 0 which is height, not width.

Using '<' instead of '>' for filtering.

The image shape attribute gives dimensions. Index 1 corresponds to width. We filter images with width greater than 30 using '>'.