Bài kiểm tra có liên kết hướng dẫn · Trung bình Cấp độ
Phục hồi lưới con người 3D với SMPL Quiz
Kiểm tra các tham số hình dạng và tư thế SMPL, HMR một hình ảnh, độ không rõ ràng của phép chiếu, các bề mặt ẩn, giới hạn quần áo, xác thực 3D và quyền riêng tư.
Câu hỏi 1 của 8
What does the basic SMPL model provide to a mesh-recovery system?
Xem trước mọi câu hỏi8 câu hỏi
Câu trả lời và giải thích xuất hiện khi bạn làm bài kiểm tra.
- What does the basic SMPL model provide to a mesh-recovery system?
- A single fixed body mesh whose pose and proportions cannot vary
- An exact clothed-surface mesh including every garment fold
- A measured 3D scan of each input person's full body
- A parametric articulated body mesh with a consistent structure
- In the guide's SMPL explanation, what do pose parameters control?
- Overall body proportions represented by shape coefficients
- The articulation of body joints and resulting mesh configuration
- Camera position and projection relative to the body
- A fixed vertex order independent of the body's motion
- What kind of input did the cited Human Mesh Recovery research use for its single-image estimate?
- A complete 3D body scan required for every prediction
- Synchronized multi-view footage required for every estimate
- One RGB image with learned body and camera inference
- A separate depth map from a range camera for every frame
- Why does a mesh fitting the visible 2D silhouette not prove exact 3D body dimensions?
- Different depth, pose and camera configurations can project similarly
- A silhouette always exposes the back of the body
- A closer camera and smaller body necessarily make different 2D images
- SMPL stores every hidden depth measurement in the image
- Which visible details are outside the basic SMPL body mesh's promised geometry?
- Body pose and learned body shape
- Loose garments and hair strands
- The model's skeletal articulation
- The mesh's shared vertex structure
- A person's arm is hidden behind the torso in one photograph. What should a reviewer infer about the recovered arm pose?
- Occlusion makes any possible arm pose equally likely
- It was observed from the camera's hidden-side view
- The mesh proves the arm was in exactly one physical position
- It is an uncertain model estimate because the arm is not directly visible
- Which test is stronger than checking only a plausible 2D overlay?
- Comparing recovered 3D pose and shape with independent reference data where available
- Reusing the same person's near-identical frames as independent test cases
- Comparing only the displayed mesh color with the photograph
- A viewer's impression that the rendering looks smooth
- Why do consent and retention controls apply to mesh-recovery outputs as well as photographs?
- A mesh is always publicly anonymous because it has no color
- Derived parameters can never be linked to the input person
- Images and inferred body measurements can reveal personal information
- A prediction automatically authorizes future reuse of the image