Feature matching between images in Computer Vision - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to create a SIFT feature detector.

Computer Vision

import cv2
sift = cv2.[1]()

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

ABRISK_create

BFeatureDetector_create

CORB_create

DSIFT_create

Attempts:

3 left

💡 Hint

Common Mistakes

Using ORB or BRISK instead of SIFT.

Trying to use a generic FeatureDetector_create method.

The correct method to create a SIFT detector in OpenCV is SIFT_create().

2fill in blank

medium

Complete the code to detect keypoints and compute descriptors from an image using SIFT.

Computer Vision

keypoints, descriptors = sift.[1](image, None)

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

AdetectAndCompute

Bdetect

Ccompute

DfindKeypoints

Attempts:

3 left

💡 Hint

Common Mistakes

Using only detect or compute separately.

Using a non-existent method like findKeypoints.

The detectAndCompute method detects keypoints and computes descriptors in one step.

3fill in blank

hard

Fix the error in the code to create a BFMatcher with L2 norm for SIFT descriptors.

Computer Vision

bf = cv2.BFMatcher([1], crossCheck=True)

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

Acv2.NORM_INF

Bcv2.NORM_HAMMING

Ccv2.NORM_L2

Dcv2.NORM_MINMAX

Attempts:

3 left

💡 Hint

Common Mistakes

Using NORM_HAMMING which is for binary descriptors like ORB.

Using incorrect norm types.

SIFT descriptors are float vectors, so the correct norm type is cv2.NORM_L2.

4fill in blank

hard

Fill both blanks to filter matches using Lowe's ratio test.

Computer Vision

good_matches = []
for m, n in matches:
    if m.[1] < [2] * n.distance:
        good_matches.append(m)

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

Adistance

B0.75

C0.5

DqueryIdx

Attempts:

3 left

💡 Hint

Common Mistakes

Using wrong attributes like queryIdx.

Using incorrect ratio values.

Lowe's ratio test compares the distance of the best match to the second best match using the distance attribute and a ratio threshold like 0.75.

5fill in blank

hard

Fill all three blanks to draw matches between two images.

Computer Vision

img_matches = cv2.drawMatches(img1, kp1, img2, kp2, [1], None, flags=[2])
cv2.imshow('Matches', img_matches)
cv2.waitKey([3])

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

Agood_matches

Bcv2.DrawMatchesFlags_NOT_DRAW_SINGLE_POINTS

C0

Dmatches

Attempts:

3 left

💡 Hint

Common Mistakes

Using all matches instead of good matches.

Not using the correct flag.

Using a non-zero waitKey value.

To draw only good matches, pass good_matches. The flag cv2.DrawMatchesFlags_NOT_DRAW_SINGLE_POINTS hides unmatched keypoints. cv2.waitKey(0) waits indefinitely for a key press.