Handwriting recognition basics in Computer Vision - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

What is the main purpose of preprocessing in handwriting recognition?

In handwriting recognition, preprocessing is an important step before feeding data to the model. What is the main goal of preprocessing?

A To increase the size of the dataset by creating new handwritten samples

B To clean and normalize the input images so the model can learn better

C To train the model faster by reducing the number of layers

D To convert the handwritten text into audio signals

💡 Hint

Think about what helps the model understand different handwriting styles better.

Preprocessing cleans and normalizes images, such as resizing and noise removal, so the model can focus on the important features.

❓ Predict Output

intermediate

Output of a simple handwriting image normalization code

What is the shape of the image after normalization in the code below?

Computer Vision

import numpy as np
from PIL import Image

img = Image.open('handwritten_sample.png').convert('L')
img_resized = img.resize((28, 28))
img_array = np.array(img_resized) / 255.0
print(img_array.shape)

A (28, 28)

B (1, 28, 28)

C (28, 28, 1)

D (784,)

💡 Hint

Look at how the image is resized and converted to an array.

The image is resized to 28x28 pixels and converted to a 2D numpy array with shape (28, 28).

❓ Model Choice

advanced

Best model type for recognizing handwritten digits

Which model type is most suitable for recognizing handwritten digits from images?

A Linear Regression

B Recurrent Neural Network (RNN)

C Convolutional Neural Network (CNN)

D K-Means Clustering

💡 Hint

Think about which model handles images and spatial patterns well.

CNNs are designed to detect patterns in images, making them ideal for handwriting recognition.

❓ Metrics

advanced

Choosing the right metric for handwriting recognition accuracy

Which metric best measures how well a handwriting recognition model correctly identifies digits?

A Perplexity

B Mean Squared Error

C Silhouette Score

D Accuracy

💡 Hint

Consider a metric that counts correct predictions over total predictions.

Accuracy measures the percentage of correct predictions, which is suitable for classification tasks like digit recognition.

🔧 Debug

expert

Why does this handwriting recognition model fail to improve accuracy?

Consider this training loop for a handwriting recognition model. Why does the accuracy stay low after many epochs?

for epoch in range(10):
    for images, labels in train_loader:
        outputs = model(images)
        loss = criterion(outputs, labels)
        optimizer.zero_grad()
        loss.backward()
        optimizer.step()
    print(f'Epoch {epoch+1}, Loss: {loss.item()}')

A The model is not set to training mode, so dropout and batch norm behave incorrectly

B The optimizer.zero_grad() is called after loss.backward(), causing gradient accumulation

C The loss is not computed using the correct labels tensor shape

D The print statement is inside the inner loop, causing too many outputs

💡 Hint

Think about what happens if the model is not told it is training.

If the model is not in training mode, layers like dropout and batch normalization do not work properly, hurting learning.