Image as numerical data (pixels, channels) in Computer Vision - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

What is the shape of the image array after loading?

Consider a color image loaded as a NumPy array with shape (height, width, channels). If the image is 128 pixels high, 256 pixels wide, and has 3 color channels (RGB), what will be the shape of the array?

Computer Vision

import numpy as np
image = np.zeros((128, 256, 3))
print(image.shape)

A (128, 256, 3)

B (3, 128, 256)

C (256, 128, 3)

D (128, 3, 256)

💡 Hint

Remember the shape format is (height, width, channels) for most image arrays.

The shape of the image array is (128, 256, 3) because the first dimension is height (rows), second is width (columns), and third is the number of color channels (RGB).

🧠 Conceptual

intermediate

Why do images have multiple channels?

Why do color images have multiple channels instead of just one?

A Because channels represent different image file formats.

B Because multiple channels store different image resolutions.

C Because channels are used to store image metadata.

D Because each channel represents a different color component like Red, Green, and Blue.

💡 Hint

Think about how colors are combined on screens.

Color images have multiple channels to represent different color components. Typically, Red, Green, and Blue channels combine to create the full color image.

❓ Hyperparameter

advanced

Choosing the right input shape for a CNN

You want to train a convolutional neural network (CNN) on 64x64 RGB images. What should be the correct input shape for the model?

A (64, 64, 3)

B (3, 64, 64)

C (64, 3, 64)

D (64, 64)

💡 Hint

Most deep learning libraries expect images with height, width, and channels in that order.

The input shape for CNNs usually follows (height, width, channels) for RGB images, so (64, 64, 3) is correct.

❓ Metrics

advanced

Calculating total pixels in a batch of images

You have a batch of 32 grayscale images, each of size 28x28 pixels. How many total pixel values are there in the batch?

A 784

B 25088

C 896

D 1024

💡 Hint

Multiply the number of images by pixels per image.

Each image has 28x28 = 784 pixels. For 32 images, total pixels = 32 * 784 = 25088.

🔧 Debug

expert

Why does this image normalization code fail?

You try to normalize an image array with pixel values from 0 to 255 by dividing by 255. But the output array remains integers. Why?

Computer Vision

import numpy as np
image = np.array([[0, 128, 255]], dtype=np.uint8)
normalized = image / 255
print(normalized)
print(normalized.dtype)

A Because dividing uint8 by int returns uint8, so decimals are lost.

B Because the division operator does not work on NumPy arrays.

C Because the image array needs to be converted to float before division.

D Because 255 is treated as zero in this context.

💡 Hint

Check the data type before and after division.

NumPy keeps the data type during division if the array is integer type, causing truncation. Converting to float before division preserves decimals.