Research map / 3D vision / Depth and stereo
FACSIMILE: Fast and Accurate Scans From an Image in Less Than a Second
David Smith, Matthew Loper, Xiaochen Hu, Paris Mavroidis, Javier Romero
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 |
|---|---|---|---|
| add normals loss terms (L1 and VGG) (ablation) | FAX depth estimator · D-FAUST depth ablation (87 subjects) | mesh distance error: 9.02 -> 6.89 | Improved |
| add depth loss term (ablation) | FAX depth estimator · D-FAUST depth ablation (87 subjects) | mesh distance error (opt back): 8.57 -> 6.66 | Improved |
| use 4 downsampling scales instead of 2 (ablation) | FAX depth estimator · D-FAUST depth ablation (87 subjects) | mesh distance error: 8.21 -> 6.89 | Improved |
| add VGG loss on normals (ablation) | FAX depth estimator · D-FAUST depth ablation (87 subjects) | mesh distance error: 7.80 -> 6.89 | Improved |
| add VGG loss on normals (ablation) | FAX depth estimator · D-FAUST depth ablation (87 subjects) | mesh distance error (opt back, pose): 3.60 -> 3.65 | No clear effect |
| add blur augmentation (ablation) | FAX depth estimator · D-FAUST depth ablation (87 subjects) | mesh distance error: 6.99 -> 6.89 | Improved |
| use 9 residual blocks instead of 5 (ablation) | FAX depth estimator · D-FAUST depth ablation (87 subjects) | mesh distance error: 6.76 -> 6.89 | No clear effect |
| use normals-only loss instead of depth-only L1 loss (ablation) | FAX depth estimator · natural test images | mesh quality | Improved |
| add adversarial loss to geometry estimation (ablation) | FAX depth estimator · geometry estimation | noise robustness | Hurt |
| optimize pose for FAX alignment evaluation | FAX · D-FAUST subject 50002 | mesh error (mm): 9.46 -> 6.56 | Improved |
| single-image FAX versus multi-image prior method | FAX · D-FAUST subject 50002 | mesh error (mm): 5.13 -> 9.46 | Hurt |
| add sharp shadows to training data (ablation) | FAX · test output | ridge artifacts | Improved |
Closest papers by meaning
- Tex2Shape: Detailed Full Human Body Geometry From a Single Image (ICCV 2019 · similarity 0.87)
- Detailed, Accurate, Human Shape Estimation From Clothed 3D Scan Sequences (CVPR 2017 · similarity 0.86)
- OmniFit: Multi-modal 3D Body Fitting via Scale-agnostic Dense Landmark Prediction (ECCV 2026 · similarity 0.86)
- TexturePose: Supervising Human Mesh Estimation With Texture Consistency (ICCV 2019 · similarity 0.86)
- Accurate 3D Body Shape Regression Using Metric and Semantic Attributes (CVPR 2022 · similarity 0.85)
- MeasureXpert: Automatic Anthropometric Measurement Extraction from Two Unregistered, Partial, Posed, and Dressed Body Scans (ICCV 2025 · similarity 0.85)
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.
