Corner detection (Harris) in Computer Vision - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to convert the image to grayscale before processing.

Computer Vision

gray = cv2.[1](img, cv2.COLOR_BGR2GRAY)

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

AcvtColor

Bresize

CGaussianBlur

Dthreshold

Attempts:

3 left

💡 Hint

Common Mistakes

Using resize instead of color conversion

Using GaussianBlur instead of color conversion

We use cv2.cvtColor with cv2.COLOR_BGR2GRAY to convert the image to grayscale.

2fill in blank

medium

Complete the code to compute the Harris corner response.

Computer Vision

dst = cv2.cornerHarris(gray, [1], 3, 0.04)

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

A1

B3

C5

D2

Attempts:

3 left

💡 Hint

Common Mistakes

Using 1 or 2 which is too small

Using 5 which is larger than typical

The block size parameter is usually set to 3 for Harris corner detection.

3fill in blank

hard

Fix the error in thresholding the Harris response to mark corners.

Computer Vision

img[dst > [1] * dst.max()] = [0, 0, 255]

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

A0.01

B0.001

C0.1

D1

Attempts:

3 left

💡 Hint

Common Mistakes

Using 1 which selects no corners

Using too small values like 0.001 which selects too many points

Using 0.1 as the threshold factor helps select strong corners by comparing to the max response.

4fill in blank

hard

Fill both blanks to normalize and convert the Harris response for visualization.

Computer Vision

dst_norm = np.empty(dst.shape, dtype=np.float32)
cv2.normalize(dst, dst_norm, [1], [2], cv2.NORM_MINMAX)

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

A0

B255

C1

D-1

Attempts:

3 left

💡 Hint

Common Mistakes

Using 255 as max which is for 8-bit images

Using negative values which are invalid here

Normalization scales values between 0 and 1 for easier visualization.

5fill in blank

hard

Fill all three blanks to create a mask of corners and mark them on the image.

Computer Vision

mask = dst_norm > [1]
img[mask] = [[2], [3], 0]

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

A0.1

B0

C255

D128

Attempts:

3 left

💡 Hint

Common Mistakes

Using 0 as threshold selects all points

Using wrong color values for marking

We threshold at 0.1, then mark corners with red color [255, 128, 0].