Research map / 3D vision / Human pose estimation
A Deeper Look into DeepCap
Marc Habermann, Weipeng Xu, Michael Zollhöfer, Gerard Pons‐Moll, Christian Theobalt
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 a second training camera view (1 -> 2) (ablation) | DeepCap · S4 | 3DPCK: 62.11 -> 93.52 | Improved |
| add a second training camera view (1 -> 2) (ablation) | DeepCap · S4 | AMVIoU: 65.11 -> 78.44 | Improved |
| add more training cameras (3 -> 7) (ablation) | DeepCap · S4 | AMVIoU: 79.75 -> 81.73 | Improved |
| use all 14 views rather than 7 (ablation) | DeepCap · S4 | AMVIoU: 81.73 -> 82.53 | Improved |
| train on 26000 frames instead of 6500 (ablation) | DeepCap · S4 | AMVIoU: 73.41 -> 82.53 | Improved |
| train on 26000 frames instead of 13000 (ablation) | DeepCap · S4 | 3DPCK: 92.25 -> 96.74 | Improved |
| add DefNet on top of PoseNet (ablation) | DeepCap · S4 | AMVIoU: 78.51 -> 82.53 | Improved |
| add DefNet on top of PoseNet (ablation) | DeepCap · S4 | 3DPCK: 96.74 -> 96.74 | No clear effect |
| weak multi-view supervised DeepCap vs supervised fine-tuned VNect | DeepCap · S1 | 3DPCK: 95.66 -> 98.43 | Improved |
| compare to HMMR (best monocular baseline) | DeepCap · S1 | MPJPE: 72.40 -> 49.11 | Improved |
| replace multi-view baseline optimization with learned network | DeepCap · S2 | 3DPCK: 89.91 -> 94.72 | Improved |
| compare to multi-view baseline MVBL | DeepCap · S1 | AMVIoU: 87.2 -> 91.74 | Hurt |
Closest papers by meaning
- DeepCap: Monocular Human Performance Capture Using Weak Supervision (CVPR 2020 · similarity 0.98)
- Volumetric performance capture from minimal camera viewpoints (ECCV 2018 · similarity 0.89)
- Deep Volumetric Video From Very Sparse Multi-View Performance Capture (ECCV 2018 · similarity 0.88)
- Capturing Humans in Motion: Temporal-Attentive 3D Human Pose and Shape Estimation From Monocular Video (CVPR 2022 · similarity 0.87)
- MonoCap: Monocular Human Motion Capture using a CNN Coupled with a Geometric Prior (TPAMI 2018 · similarity 0.87)
- Motions as Queries: One-Stage Multi-Person Holistic Human Motion Capture (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.
