Research map / 3D vision / Face and shape
Self-supervised 3D Representation Learning of Dressed Humans from Social Media Videos
Yasamin Jafarian, Hyun Soo Park
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 joint surface-normal loss L_s to depth loss (ablation) | L_z model · Tang et al. dataset (ablation, tested without adaptation) | depth error (cm): 5.66 -> 5.11 | Improved |
| add joint surface-normal loss L_s to depth loss (ablation) | L_z model · Tang et al. dataset (ablation, tested without adaptation) | normal error (deg): 34.24 -> 29.99 | Improved |
| add self-supervised warping loss L_w with TikTok videos (ablation) | L_z + L_s · Tang et al. dataset (ablation, tested without adaptation) | depth error (cm): 5.11 -> 4.89 | Improved |
| add self-supervised warping loss L_w with TikTok videos (ablation) | L_z + L_s · Tang et al. dataset (ablation, tested without adaptation) | normal error (deg): 29.99 -> 29.36 | Improved |
| add self-supervised warping loss L_w with TikTok videos (ablation) | L_z + L_s · Tang et al. dataset (ablation, tested without adaptation) | reconstruction error (cm): 4.66 -> 4.46 | Improved |
| use affine instead of rigid transformation warping | Ours · Vlasic et al. dataset | depth error (dataset scale): 16.9 -> 16.8 | No clear effect |
| use affine instead of rigid transformation warping | Ours · Tang et al. dataset | depth error (cm): 5.0 -> 4.8 | Improved |
| add photometric consistency loss L_p | Ours (affine) · RenderPeople dataset | depth error (cm): 3.2 -> 2.9 | Improved |
| add photometric consistency loss L_p | Ours (affine) · Tang et al. dataset | depth error (cm): 4.8 -> 5.1 | Hurt |
| add photometric consistency loss L_p | Ours (affine) · Vlasic et al. dataset | depth error (dataset scale): 16.8 -> 15.9 | Improved |
Closest papers by meaning
- Learning High Fidelity Depths of Dressed Humans by Watching Social Media Dance Videos (CVPR 2021 · similarity 0.97)
- Self-Supervised Human Depth Estimation From Monocular Videos (CVPR 2020 · similarity 0.89)
- Learning Realistic Human Reposing Using Cyclic Self-Supervision With 3D Shape, Pose, and Appearance Consistency (ICCV 2021 · similarity 0.89)
- VideoMDM: Towards 3D Human Motion Generation From 2D Supervision (NeurIPS 2026 · similarity 0.87)
- Self-Supervised 3D Human Mesh Recovery from a Single Image with Uncertainty-Aware Learning (AAAI 2024 · similarity 0.87)
- MultiPly: Reconstruction of Multiple People from Monocular Video in the Wild (CVPR 2024 · similarity 0.87)
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.
