Research map / 3D vision / Face and shape
Image-Guided Shape-from-Template Using Mesh Inextensibility Constraints
Thuy Tran, Ruochen Chen, Shaifali Parashar
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 image gradient loss (ablation) | Ours · phi-SfT real (avg) | chamfer distance x1e4: 3.95 -> 3.91 | Improved |
| add adaptive data loss (ablation) | Ours · phi-SfT real (avg) | chamfer distance x1e4: 4.98 -> 3.91 | Improved |
| add silhouette loss (ablation) | Ours · phi-SfT real (avg) | chamfer distance x1e4: 3.97 -> 3.91 | Improved |
| use large (12-layer, 512-width) instead of base network (ablation) | Ours · phi-SfT real R7 | chamfer distance x1e4: 4.62 -> 6.00 | Hurt |
| use large instead of base network (ablation) | Ours · phi-SfT real R2 | chamfer distance x1e4: 1.30 -> 0.96 | Improved |
| use small (4-layer, 64-width) instead of base network (ablation) | Ours · phi-SfT real R4 | chamfer distance x1e4: 8.15 -> 15.85 | Hurt |
| use small instead of base network (ablation) | Ours · phi-SfT real R2 | chamfer distance x1e4: 1.30 -> 2.49 | Hurt |
| set sigma=0.1 in adaptive data loss exponential weight (ablation) | Ours · phi-SfT real (avg) | chamfer distance x1e4: 3.91 -> 28.86 | Hurt |
| set alpha=0.1 in adaptive data loss (ablation) | Ours · phi-SfT real (avg) | chamfer distance x1e4: 3.91 -> 18.89 | Hurt |
| increase per-frame iterations from 200 to 250 (ablation) | Ours · Kinect Paper | depth RMSE (mm): 4.01 -> 3.99 | No clear effect |
| use window-wise optimization (w=3) with temporal loss (ablation) | Ours · Kinect Paper | depth RMSE (mm): 4.01 -> 3.57 | Improved |
| use adaptive stopping per frame (tolerance 1e-6) (ablation) | Ours · Kinect Paper | depth RMSE (mm): 4.01 -> 3.64 | Improved |
Closest papers by meaning
- f-SfT: Shape-From-Template With a Physics-Based Deformation Model (CVPR 2022 · similarity 0.91)
- A Stable Analytical Framework for Isometric Shape-from-Template by Surface Integration (TPAMI 2016 · similarity 0.91)
- Physics-guided Shape-from-Template: Monocular Video Perception through Neural Surrogate Models (CVPR 2024 · similarity 0.90)
- Unsupervised Template-assisted Point Cloud Shape Correspondence Network (CVPR 2024 · similarity 0.85)
- PyTorchGeoNodes: Enabling Differentiable Shape Programs for 3D Shape Reconstruction (CVPR 2025 · similarity 0.84)
- SF3D: Stable Fast 3D Mesh Reconstruction with UV-unwrapping and Illumination Disentanglement (CVPR 2025 · similarity 0.84)
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.
