Research map / 3D vision / Depth and stereo
3DFG-PIFu: 3D Feature Grids for Human Digitization from Sparse Views
Kennard Yanting Chan, Fayao Liu, Guosheng Lin, Chuan Sheng Foo, Weisi Lin
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 3D visual hull feature grid GM to single-view PIFu (ablation) | PIFu · THuman2.0 | Chamfer distance (10^-5): 26.97 -> 6.626 | Improved |
| add front-normal grid GN alongside GM (ablation) | PIFu + GM · THuman2.0 | Chamfer distance (10^-5): 6.626 -> 6.007 | Improved |
| add view visibility grid GV to 2nd stage (ablation) | 3DFG-PIFu 2nd stage · THuman2.0 visible regions | Chamfer distance (10^-5): 4.036 -> 3.891 | Improved |
| add iterative 2nd stage refinement (ablation) | 3DFG-PIFu · THuman2.0 | Chamfer distance (10^-5): 6.007 -> 5.796 | Improved |
| add iterative 2nd stage refinement (ablation) | 3DFG-PIFu · BUFF | Chamfer distance (10^2): 2.386 -> 2.509 | Hurt |
| use SDF-based SMPL-X features GX instead of S-PIFu features (ablation) | single-view PIFu · THuman2.0 | Chamfer distance (10^-5): 4.488 -> 4.352 | Improved |
| combine GX with PaMIR voxel-aligned features (ablation) | single-view PIFu · THuman2.0 | Chamfer distance (10^-5): 4.104 -> 3.970 | Improved |
| combine GX with voxel-aligned features (ablation) | single-view PIFu · BUFF | Chamfer distance (10^2): 8.037 -> 8.012 | Mixed |
| use ground-truth SMPL-X mesh and high-res images in 3DFG-PIFu | 3DFG-PIFu · BUFF | Chamfer distance (10^2): 2.509 -> 3.412 | Mixed |
| replace point-wise multi-view fusion with 3D feature grids | 3DFG-PIFu vs SeSDF · THuman2.0 | Chamfer distance (10^-5): 6.302 -> 5.796 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SeSDF — SeSDF: Self-Evolved Signed Distance Field for Implicit 3D Clothed Human Reconstruction (CVPR 2023) · better on 5 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Geo-PIFu: Geometry and Pixel Aligned Implicit Functions for Single-view Human Reconstruction (NeurIPS 2020 · similarity 0.91)
- SIFU: Side-view Conditioned Implicit Function for Real-world Usable Clothed Human Reconstruction (CVPR 2024 · similarity 0.87)
- Fine Structure-Aware Sampling: A New Sampling Training Scheme for Pixel-Aligned Implicit Models in Single-View Human Reconstruction (AAAI 2024 · similarity 0.86)
- RoGSplat: Learning Robust Generalizable Human Gaussian Splatting from Sparse Multi-View Images (CVPR 2025 · similarity 0.86)
- PF3plat: Pose-Free Feed-Forward 3D Gaussian Splatting for Novel View Synthesis (ICML 2025 · similarity 0.85)
- VS: Reconstructing Clothed 3D Human from Single Image via Vertex Shift (CVPR 2024 · 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.
