引導式測驗 · 中等 等級
面部标志检测测验
測試臉部標誌輸出、MediaPipe 配置、幾何限制、特定於任務的評估以及臉部資料的隱私意識使用。
問題 1 的 8
Which output is the direct purpose of facial landmark detection?
預覽每個問題8 問題
當您參加測驗時,就會出現答案和解釋。
- Which output is the direct purpose of facial landmark detection?
- Proof of the person’s intention
- Estimated coordinates for selected points on a detected face
- A verified identity for the person
- A diagnosis from facial appearance
- In the documented MediaPipe Face Landmarker model bundle, how many 3D face landmarks are estimated?
- 52
- 21
- 106
- 478
- What does a facial transformation matrix support in the documented task?
- Mapping a canonical face model to detected landmarks for effects rendering
- Determining the subject’s legal identity
- Converting landmark points into a medical diagnosis
- Measuring the user’s consent from facial pose
- A filter jitters when a person turns sideways. What should the team measure?
- Whether the output file uses a standard image extension
- The model’s performance on unrelated object classes
- Landmark and tracking stability across the target poses and conditions
- Only the number of coordinates emitted in a frontal image
- Which claim is supported by landmark coordinates alone?
- They prove a person is happy
- They confirm the person’s identity
- They show consent to camera processing
- They provide estimated face geometry useful for alignment or effects
- Which image conditions can make face landmarks less accurate?
- The interface labels a point as normalized
- Occlusion, lighting, pose, or camera quality can affect image evidence
- The model outputs a transformation matrix
- The app stores no face images
- How should optional blendshape output be treated?
- As a substitute for face detection
- As proof that the model knows the user’s intent
- As a configured model output used by downstream applications, not an identity result
- As a guaranteed emotion label with universal meaning
- Which privacy practice fits a camera feature using facial landmarks?
- Explain face processing, minimize storage, and provide a way to disable it
- Keep the camera indicator hidden to reduce distraction
- Store every face crop indefinitely for convenience
- Treat local processing as a reason to omit user controls