Guide-linked quiz · Medium Level

SIFT and ORB Feature Detection Quiz

Compare SIFT and ORB keypoints and descriptors, choose compatible distance metrics, and filter tentative matches with geometric checks.

Question 1 of 8

Around a detected keypoint, what information does a local image descriptor represent?

Preview every question8 questions

Answers and explanations appear as you take the quiz.

  1. Around a detected keypoint, what information does a local image descriptor represent?
    • The average color distribution of the entire image
    • The camera’s absolute GPS position
    • The appearance around a detected keypoint
    • The full 3D geometry of the scene
  2. Which descriptor type is associated with ORB?
    • Binary
    • A floating-point gradient histogram like SIFT
    • A complete grayscale image patch
    • A text embedding of the object category
  3. Which distance is commonly used to compare SIFT descriptors?
    • Hamming distance over binary strings
    • Cosine distance over normalized binary codes
    • Edit distance over keypoint coordinates
    • L2 / Euclidean distance
  4. Which distance is commonly used for standard ORB binary descriptors?
    • L2 distance over floating-point vectors
    • Hamming distance
    • Cosine distance over token embeddings
    • A ratio of image brightness values
  5. What does the nearest-neighbor ratio test help detect?
    • Whether the camera has been calibrated
    • Whether keypoints occur on moving objects
    • Ambiguous matches where the nearest two descriptor candidates are too similar
    • Whether the image contains a bright region
  6. Why should descriptor matches be checked against a geometric model?
    • Some visually similar descriptors are false correspondences
    • Descriptor similarity alone guarantees a geometrically valid pair
    • Every descriptor match is a proven physical point
    • Geometric checks make image overlap unnecessary
  7. What may be needed when ORB uses WTA_K of 3 or 4 in OpenCV?
    • L2 distance because all ORB descriptors are floating point
    • The Hamming2 distance metric
    • The SIFT descriptor converted to grayscale
    • A ratio test instead of any descriptor metric
  8. How should a team compare SIFT and ORB for an application?
    • Count raw matches on one easy image only
    • Choose whichever detector returns the most keypoints
    • Assume one method leads on every scene
    • Measure quality and runtime on representative images and target hardware

Keep testing yourself

More quizzes picked for your level.

View all quizzes