Super-resolution basics in Computer Vision - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to load a low-resolution image using OpenCV.

Computer Vision

import cv2
low_res_img = cv2.imread([1])
print(low_res_img.shape)

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

A'low_res_image.jpg'

B'model.pth'

C'output.png'

D'high_res_image.jpg'

Attempts:

3 left

💡 Hint

Common Mistakes

Using a high-resolution image filename instead of low-resolution.

We load the low-resolution image file named 'low_res_image.jpg' to start super-resolution.

2fill in blank

medium

Complete the code to resize the low-resolution image to a higher resolution using OpenCV.

Computer Vision

high_res_img = cv2.resize(low_res_img, [1], interpolation=cv2.INTER_CUBIC)
print(high_res_img.shape)

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

A(256, 256)

B(64, 64)

C(16, 16)

D(8, 8)

Attempts:

3 left

💡 Hint

Common Mistakes

Choosing smaller or same size dimensions, which does not increase resolution.

We resize the image to a larger size (256x256) to simulate super-resolution output.

3fill in blank

hard

Fix the error in the code to define a simple super-resolution model using PyTorch.

Computer Vision

import torch.nn as nn
class SimpleSR(nn.Module):
    def __init__(self):
        super(SimpleSR, self).__init__()
        self.conv1 = nn.Conv2d(3, 64, kernel_size=[1], padding=1)
    def forward(self, x):
        return self.conv1(x)

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

A1

B5

C3

D7

Attempts:

3 left

💡 Hint

Common Mistakes

Using kernel size 1 or 7 which changes feature extraction behavior.

The kernel size should be 3 to keep the output size consistent with padding=1.

4fill in blank

hard

Fill both blanks to complete the training loop snippet for super-resolution model.

Computer Vision

for epoch in range(num_epochs):
    for data in dataloader:
        inputs, targets = data
        optimizer.zero_grad()
        outputs = model([1])
        loss = criterion(outputs, [2])
        loss.backward()
        optimizer.step()

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

Ainputs

Btargets

Coutputs

Dinputs.detach()

Attempts:

3 left

💡 Hint

Common Mistakes

Passing targets to the model or comparing outputs with inputs.

We pass inputs to the model and compare outputs with targets to compute loss.

5fill in blank

hard

Fill all three blanks to create a dictionary comprehension that maps image names to their super-resolution outputs.

Computer Vision

sr_results = [1]: model(images[[2]].unsqueeze(0)) for [3] in images.keys()}

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

Aname

Dimg

Attempts:

3 left

💡 Hint

Common Mistakes

Using different variable names inconsistently causing errors.

We use 'name' as the key and loop variable, and access images by 'name' to get the input tensor.