Research map / 3D vision / Human pose estimation
Explaining Human Preferences via Metrics for Structured 3D Reconstruction
Jack Langerman, Denys Rozumnyi, Yuzhong Huang, Dmytro Mishkin
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use edge F1 instead of WED with pre-registration | wireframe metric · unit tests | tests passed: 8/17 -> 14/17 | Improved |
| use corner F1 instead of Hausdorff distance | wireframe metric · unit tests | tests passed: 6/17 -> 12/17 | Improved |
| use edge recall instead of edge precision | wireframe metric · unit tests | tests passed: 12/17 -> 13/17 | Improved |
| use Jaccard distance instead of WED pre-registration | wireframe metric · agreement with human preference | agreement | Improved |
| use recall metrics instead of precision metrics | wireframe metric · agreement with human preference | agreement | Improved |
| restrict agreement to decisive pairs instead of all pairs | annotator agreement · wireframe ranking | agreement (%): 80% -> 91% | Improved |
| restrict agreement to within annotator clusters | annotator agreement · wireframe ranking | agreement (%): 80% -> 85% | Improved |
| use VLMs as metrics instead of expert raters | VLMs · wireframe ranking | agreement with humans | No clear effect |
| use spectral L1 distance instead of human-aligned metrics | spectral L1 · unit tests and human agreement | tests passed: 13/17 | Mixed |
| use Chamfer edge distance | Chamfer edge · unit tests and human agreement | tests passed: 8/17 | Mixed |
Closest papers by meaning
- SRAM: Shape-Realism Alignment Metric for No Reference 3D Shape Evaluation (AAAI 2026 · similarity 0.86)
- Learning 3D Shape Fidelity Metric from Real-world Distortions (CVPR 2026 · similarity 0.86)
- Textured Geometry Evaluation: Perceptual 3D Textured Shape Metric via 3D Latent-Geometry Network (AAAI 2026 · similarity 0.85)
- Evaluating Multiview Object Consistency in Humans and Image Models (NeurIPS 2024 · similarity 0.83)
- Gen3DEval: Using vLLMs for Automatic Evaluation of Generated 3D Objects (CVPR 2025 · similarity 0.83)
- Metric from Human: Zero-shot Monocular Metric Depth Estimation via Test-time Adaptation (NeurIPS 2024 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
