Research map / Image generation / Super-resolution
Super-Resolution 3D Human Shape from a Single Low-Resolution Image
Marco Pesavento, Marco Volino, Adrian Hilton
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 |
|---|---|---|---|
| train all modules end to end (ablation) | SuRS · THuman2.0 (LR input 256) | Chamfer distance (cm): 1.375 -> 0.931 | Improved |
| train all modules end to end instead of feature extractor then MLPs (ablation) | SuRS · THuman2.0 (LR input 256) | Chamfer distance (cm): 1.261 -> 0.931 | Improved |
| train all modules end to end instead of extractor+MR-MLP then SR-MLP (ablation) | SuRS · THuman2.0 (LR input 256) | Chamfer distance (cm): 1.035 -> 0.931 | Improved |
| add U-Net high-resolution feature extractor with 2D reconstruction loss (ablation) | SuRS · THuman2.0 (LR input 256) | Chamfer distance (cm): 1.090 -> 0.931 | Improved |
| add SR-MLP stage on top of MR-MLP (ablation) | SuRS · THuman2.0 (LR input 256) | Chamfer distance (cm): 1.044 -> 0.931 | Improved |
| add MR-MLP conditioning to SR-MLP (ablation) | SuRS · THuman2.0 (LR input 256) | Chamfer distance (cm): 1.039 -> 0.931 | Improved |
| add difference loss Ldiff (ablation) | SuRS · THuman2.0 (LR input 256) | Chamfer distance (cm): 1.020 -> 0.931 | Improved |
| add difference loss Ldiff (ablation) | SuRS · 3D People | Chamfer distance (cm): 1.127 -> 1.057 | Improved |
| decimate LR ground-truth shape to 1000 faces instead of 100000 (ablation) | SuRS · THuman2.0 (LR input 256) | Chamfer distance (cm): 0.942 -> 0.931 | Improved |
| decimate LR ground-truth shape to 1000 faces instead of 100000 (ablation) | SuRS · 3D People | Chamfer distance (cm): 1.147 -> 1.057 | Improved |
| decimate LR ground-truth shape to 1000 faces (ablation) | SuRS · THuman2.0 (LR input 256) | P2S (cm): 1.116 -> 1.151 | No clear effect |
| use SuRS instead of PIFuHD with normal maps | SuRS · THuman2.0 (LR input 256) | P2S (cm): 1.046 -> 1.151 | Hurt |
Closest papers by meaning
- High-Fidelity 3D Human Digitization From Single 2K Resolution Images (CVPR 2023 · similarity 0.89)
- Combining Implicit Function Learning and Parametric Models for 3D Human Reconstruction (ECCV 2020 · similarity 0.88)
- Complete 3D Human Reconstruction From a Single Incomplete Image (CVPR 2023 · similarity 0.87)
- MultiGO: Towards Multi-level Geometry Learning for Monocular 3D Textured Human Reconstruction (CVPR 2025 · similarity 0.87)
- HiLo: Detailed and Robust 3D Clothed Human Reconstruction with High-and Low-Frequency Information of Parametric Models (CVPR 2024 · similarity 0.86)
- VS: Reconstructing Clothed 3D Human from Single Image via Vertex Shift (CVPR 2024 · similarity 0.86)
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.
