Guide-linked quiz · Medium Level

Hand Pose Estimation and Hand Tracking Quiz

Tests landmarks, video tracking, gesture interpretation, failure modes, and human-centered validation of hand pose systems and apps.

Question 1 of 8

What does MediaPipe Hand Landmarker output for each detected hand?

Preview every question8 questions

Answers and explanations appear as you take the quiz.

  1. What does MediaPipe Hand Landmarker output for each detected hand?
    • A guarantee of the person’s intended gesture
    • A hand image with all occlusion removed
    • A set of 21 hand landmarks, including key hand points
    • A complete sign-language translation for every frame
  2. How does hand-pose estimation differ from hand tracking?
    • Estimation finds keypoints in an image; tracking associates detections over time across frames
    • Estimation identifies user intent while tracking measures emotion
    • Tracking works only on still images
    • They are identical terms for the same one-frame result
  3. Before landmarks control a gesture interface, what must the application define?
    • A wrist point that infers every hand action
    • Patterns mapped to actions, with user feedback
    • Coordinates that automatically encode the user’s command
    • A universal gesture vocabulary supplied by the landmark model
  4. When two visible hands cross in a tracked sequence, which identity error can occur?
    • The hand keypoints can become noisier while the same track IDs remain attached
    • The tracker may temporarily lose one hand but keep its last identity until detection resumes
    • The gesture classifier may confuse a pose while the hand-to-track association remains stable
    • The tracker can swap which physical hand its two track IDs refer to
  5. Why test a live gesture pipeline at its intended frame rate?
    • Latency and frame sampling can affect tracking continuity and action timing
    • Camera frame rate sets temporal sampling, but cannot affect action timing once points are detected
    • Display refresh rate alone determines tracking latency regardless of camera input
    • The same smoothing window preserves timing at every capture frame rate
  6. What does a landmark model alone not provide?
    • A set of landmarks in image coordinates
    • Detection results for visible hand poses
    • A guaranteed interpretation of what a person intends
    • The estimated locations of hand keypoints
  7. Which evaluation is most useful for a hand-controlled app?
    • Check whether the camera lens is a wide-angle model
    • Measure localization error, misses, identity switches, latency, and user comfort
    • Count only the number of keypoints in the model
    • Measure training accuracy on one clean video
  8. For a high-impact gesture action, which safeguards should the app provide?
    • Smooth a pose across frames and show feedback, then trigger automatically without a cancel path
    • Require a one-time setup confirmation and trigger later actions from a single high-confidence frame
    • Provide pause and undo controls, but allow a single-frame gesture to trigger a consequential action
    • Require stable, user-confirmed input and provide pause or undo controls

Keep testing yourself

More quizzes picked for your level.

View all quizzes