Feature extraction approach in Computer Vision - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Understanding Feature Extraction in Image Classification

Which of the following best describes the purpose of feature extraction in image classification?

A To reduce the image size by cropping and resizing before classification

B To randomly shuffle image pixels to improve model robustness

C To increase the number of pixels in the image for better resolution

D To transform raw image data into a set of meaningful values that represent important characteristics

💡 Hint

Think about how we simplify complex images into useful information for the model.

Feature extraction transforms raw images into meaningful values that capture important patterns, making it easier for models to learn.

❓ Predict Output

intermediate

Output of Feature Extraction Using SIFT

What is the output type of the following Python code using OpenCV's SIFT feature extractor?

Computer Vision

import cv2
img = cv2.imread('image.jpg', cv2.IMREAD_GRAYSCALE)
sift = cv2.SIFT_create()
keypoints, descriptors = sift.detectAndCompute(img, None)
print(type(descriptors))

A <class 'numpy.ndarray'>

B <class 'list'>

C <class 'NoneType'>

D <class 'dict'>

💡 Hint

Descriptors are numerical arrays representing features.

The SIFT extractor returns descriptors as a numpy array, where each row corresponds to a feature vector.

❓ Model Choice

advanced

Choosing a Model for Feature Extraction

You want to extract features from images to use in a simple classifier. Which model is best suited for automatic feature extraction without manual design?

A A decision tree classifier

B A linear regression model

C A pretrained convolutional neural network like ResNet

D A k-nearest neighbors algorithm

💡 Hint

Consider models that learn hierarchical image features automatically.

Pretrained CNNs like ResNet automatically learn rich image features useful for many tasks, unlike simpler models that need manual features.

❓ Hyperparameter

advanced

Effect of Number of Features in Feature Extraction

When using a feature extractor like ORB, increasing the number of features parameter will most likely:

A Decrease the computation time and reduce feature quality

B Increase the number of detected keypoints and increase computation time

C Have no effect on the number of features detected

D Cause the extractor to ignore edges and focus on colors

💡 Hint

Think about what happens when you ask for more features from the extractor.

Increasing the number of features lets the extractor find more keypoints but requires more processing time.

❓ Metrics

expert

Evaluating Feature Extraction Quality

You extracted features from images and trained a classifier. Which metric best helps evaluate if the features separate classes well?

A Silhouette score measuring cluster separation of features

B Mean squared error between features

C Number of features extracted per image

D Accuracy of the classifier on a test set

💡 Hint

Think about a metric that measures how well features group similar samples apart from others.

Silhouette score measures how well features cluster by class, indicating good feature separation.