Research map / 3D vision / Face and shape
Combining Implicit Function Learning and Parametric Models for 3D Human Reconstruction
Bharat Lal Bhatnagar, Cristian Sminchisescu, Christian Theobalt, Gerard Pons-Moll
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 inner body surface prediction to outer-only registration (ablation) | IP-Net registration · Dressed human scans, 50 test (SMPL+D registration, vertex error cm) | vertex-to-vertex error (cm), outer: 11.84 -> 11.54 | Improved |
| add part correspondences to outer+inner registration (ablation) | IP-Net registration · Dressed human scans, 50 test (SMPL+D registration, vertex error cm) | vertex-to-vertex error (cm), outer: 11.54 -> 3.67 | Improved |
| add part correspondences to outer+inner registration (ablation) | IP-Net registration · Dressed human scans, 50 test (SMPL+D registration, vertex error cm) | vertex-to-vertex error (cm), inner: 11.14 -> 3.32 | Improved |
| register SMPL+D to IP-Net predictions instead of IF-Net reconstruction | SMPL+D registration · Dressed human scans, 50 test (SMPL+D registration, vertex error cm) | vertex-to-vertex error (cm): 13.88 -> 3.67 | Improved |
| register to implicit predictions instead of regressing SMPL+D parameters | SMPL+D fitting · Dressed human scans, 50 test (SMPL+D registration, vertex error cm) | vertex-to-vertex error (cm): 32.45 -> 3.67 | Improved |
| use IP-Net predictions instead of IP-Net correspondences only on single view (ablation) | SMPL+D registration · Single-view point cloud | vertex-to-vertex error (cm), outer: 14.43 -> 5.11 | Improved |
| jointly predict inner and outer surface instead of two independent networks (ablation) | IP-Net · Inner/outer surface inter-penetration | intersecting face area (mm^2): 2000.71mm^2 -> 0.65mm^2 | Improved |
| use IP-Net part labels for scan registration instead of registering sparse point cloud | SMPL+D registration · Scan registration | vertex-to-vertex error (cm), outer: 13.93 -> 3.67 | Improved |
| use IP-Net instead of IF-Net for outer surface reconstruction | implicit reconstruction · Outer surface reconstruction | vertex-to-surface error (mm): 4.86 -> 4.95 | No clear effect |
Closest papers by meaning
- SHARP: Shape-Aware Reconstruction of People in Loose Clothing (IJCV 2022 · similarity 0.91)
- Multi-Person Implicit Reconstruction From a Single Image (CVPR 2021 · similarity 0.91)
- ECON: Explicit Clothed Humans Optimized via Normal Integration (CVPR 2023 · similarity 0.90)
- SCALE: Modeling Clothed Humans with a Surface Codec of Articulated Local Elements (CVPR 2021 · similarity 0.90)
- HISR: Hybrid Implicit Surface Representation for Photorealistic 3D Human Reconstruction (AAAI 2024 · similarity 0.90)
- VS: Reconstructing Clothed 3D Human from Single Image via Vertex Shift (CVPR 2024 · similarity 0.89)
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.
