Privacy considerations in Computer Vision - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to load an image while respecting privacy by blurring faces.

Computer Vision

import cv2
image = cv2.imread('input.jpg')
face_cascade = cv2.CascadeClassifier('haarcascade_frontalface_default.xml')
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
faces = face_cascade.detectMultiScale(gray, 1.1, 4)
for (x, y, w, h) in faces:
    face_region = image[y:y+h, x:x+w]
    blurred_face = cv2.[1](face_region, (99, 99), 30)
    image[y:y+h, x:x+w] = blurred_face
cv2.imwrite('output.jpg', image)

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

Athreshold

Bresize

CcvtColor

DGaussianBlur

Attempts:

3 left

💡 Hint

Common Mistakes

Using resize instead of blur will not obscure the face.

Using cvtColor or threshold does not blur the image.

The GaussianBlur function blurs the detected face region to protect privacy by obscuring facial details.

2fill in blank

medium

Complete the code to anonymize faces by pixelating them.

Computer Vision

import cv2
image = cv2.imread('input.jpg')
face_cascade = cv2.CascadeClassifier('haarcascade_frontalface_default.xml')
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
faces = face_cascade.detectMultiScale(gray, 1.1, 4)
for (x, y, w, h) in faces:
    face_region = image[y:y+h, x:x+w]
    small = cv2.resize(face_region, ([1], [1]), interpolation=cv2.INTER_LINEAR)
    pixelated_face = cv2.resize(small, (w, h), interpolation=cv2.INTER_NEAREST)
    image[y:y+h, x:x+w] = pixelated_face
cv2.imwrite('output_pixelated.jpg', image)

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

A10

B5

C50

D100

Attempts:

3 left

💡 Hint

Common Mistakes

Using a large size like 100 will not pixelate the face.

Choosing 50 or 10 may not anonymize enough.

Using a small size like 5x5 pixels for resizing creates a pixelated effect that anonymizes the face.

3fill in blank

hard

Fix the error in the code that attempts to remove metadata from an image to protect privacy.

Computer Vision

from PIL import Image
image = Image.open('photo.jpg')
if 'exif' in image.info:
    image.info.pop([1])
image.save('photo_no_metadata.jpg')

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

A'info'

B'metadata'

C'exif'

D'data'

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'metadata' or 'info' as keys will cause a KeyError.

Not removing 'exif' leaves metadata in the saved image.

The key to remove in the image info dictionary to delete metadata is 'exif'.

4fill in blank

hard

Fill both blanks to create a function that detects faces and replaces them with black rectangles to protect privacy.

Computer Vision

import cv2
def anonymize_faces(image_path):
    image = cv2.imread(image_path)
    face_cascade = cv2.CascadeClassifier('haarcascade_frontalface_default.xml')
    gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    faces = face_cascade.detectMultiScale(gray, 1.1, 4)
    for (x, y, w, h) in faces:
        cv2.[1](image, (x, y), (x + w, y + h), [2], -1)
    return image

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

Arectangle

B(0, 0, 0)

C(255, 255, 255)

Dcircle

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'circle' will not cover the entire face properly.

Using white color (255,255,255) may not anonymize well on light backgrounds.

The rectangle function draws black rectangles (color (0,0,0)) over faces to anonymize them.

5fill in blank

hard

Fill all three blanks to create a dictionary comprehension that filters out sensitive keys from metadata.

Computer Vision

metadata = {'name': 'Alice', 'age': 30, 'ssn': '123-45-6789', 'email': '[e-mail]'}
safe_metadata = {k: v for k, v in metadata.items() if k not in [1] and v != [2] and k != [3]

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

A['ssn', 'email']

BNone

C'age'

D['password', 'token']

Attempts:

3 left

💡 Hint

Common Mistakes

Using wrong keys in the list will not filter sensitive data.

Using wrong value for comparison will not filter correctly.

We exclude keys 'ssn' and 'email' in the list, filter out values equal to None, and exclude key 'age' to protect privacy.