Research map / 3D vision / Face and shape
Monocular Real-Time Volumetric Performance Capture
Ruilong Li, Yuliang Xiu, Shunsuke Saito, Zeng Huang, Kyle Olsewski, Hao Li
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 point-ohem sampling (ablation) | PIFu-based real-time model · RenderPeople | Chamfer distance: 1.561 -> 1.397 | Improved |
| add point-ohem sampling (ablation) | PIFu-based real-time model · BUFF | P2S distance (top-10 worst): 9.556 -> 7.092 | Improved |
| add item-ohem on top of point-ohem (ablation) | Ours+P · top-10 worst cases, BUFF | Chamfer* and P2S*: 7.092 -> 5.874 | Improved |
| add item-ohem on top of point-ohem (ablation) | Ours+P · RenderPeople | Chamfer distance (average): 1.397 -> 1.431 | Hurt |
| replace stacked hourglass with HRNet encoder (ablation) | HG-MLP · RenderPeople | Chamfer distance: 1.684 -> 1.602 | Improved |
| replace stacked hourglass with HRNet encoder (ablation) | HG-MLP · Runtime | fps: 12 -> 14 | Improved |
| add conditional batch norm (CBN) (ablation) | HRNet-MLP · RenderPeople | Chamfer distance: 1.602 -> 1.584 | Improved |
| add SoftZ depth representation (ablation) | HRNet-CBN · RenderPeople | Chamfer distance: 1.584 -> 1.561 | Improved |
| add SoftZ depth representation (ablation) | HRNet-CBN · BUFF | P2S distance: 3.585 -> 3.613 | Hurt |
| use hierarchical surface localization instead of brute force (ablation) | PIFu inference · reconstruction time | seconds per frame: 30 seconds to 0.14 seconds | Improved |
| add direct mesh-free rendering (ablation) | hierarchical localization · novel-view rendering | seconds per frame: 0.14 seconds to 0.06 seconds | Improved |
| add binarized octree localization instead of proposed method (ablation) | brute-force baseline · reconstruction | completeness | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- VIBE — VIBE: Video Inference for Human Body Pose and Shape Estimation (CVPR 2020) · better on 2 of 2 shared results
- DeepHuman — DeepHuman: 3D Human Reconstruction From a Single Image (ICCV 2019) · better on 2 of 2 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
- Deep Volumetric Video From Very Sparse Multi-View Performance Capture (ECCV 2018 · similarity 0.87)
- Function4D: Real-Time Human Volumetric Capture From Very Sparse Consumer RGBD Sensors (CVPR 2021 · similarity 0.87)
- Geo-PIFu: Geometry and Pixel Aligned Implicit Functions for Single-view Human Reconstruction (NeurIPS 2020 · similarity 0.87)
- Link to the Past: Temporal Propagation for Fast 3D Human Reconstruction from Monocular Video (CVPR 2025 · 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)
- ODHSR: Online Dense 3D Reconstruction of Humans and Scenes from Monocular Videos (CVPR 2025 · 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.
