Research map / 3D vision / Human pose estimation
Pro-Pose: Unpaired Full-Body Portrait Synthesis via Canonical UV Maps
Sandeep Mishra, Yasamin Jafarian, Andreas Lugmayr, Yingwei Li, Varsha Ramakrishnan, Srivatsan Varadharajan, Alan Bovik, Ira Kemelmacher-Shlizerman
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 unpaired single-image data to paired training (ablation) | Pro-Pose · WPose | FaceSim: 0.4959 -> 0.5571 | Improved |
| add paired data to unpaired training (ablation) | Pro-Pose · WPose | FaceSim: 0.3805 -> 0.5571 | Improved |
| add unpaired data to paired training (ablation) | Pro-Pose · WPose | M-PSNR: 18.30 -> 19.95 | Improved |
| add unpaired data to paired training (ablation) | Pro-Pose · DeepFashion (in-domain) | FaceSim: 0.6255 -> 0.6047 | Hurt |
| add paired data to unpaired training (ablation) | Pro-Pose · DeepFashion (in-domain) | PSNR: 15.66 -> 19.36 | Improved |
| use donor-based UV reposing instead of random patch masking (ablation) | Pro-Pose · WPose | FaceSim: 6.75% | Improved |
| add UV texture condition to paired-only training (ablation) | Pro-Pose (paired only) · WPose | FaceSim: 0.4721 -> 0.4959 | Improved |
| add UV texture condition to paired-only training (ablation) | Pro-Pose (paired only) · WPose | HPSv3: 6.98 -> 7.30 | Improved |
| add test-time few-shot personalization (ablation) | Pro-Pose · identity preservation | FaceSim: 18.3% | Improved |
| apply VTO on Pro-Pose avatar instead of original image | VTO · WPose user study | user preference: 28.02% -> 71.98% | Improved |
| apply VTO on Pro-Pose avatar instead of original image | VTO · WPose user study | garment error: 31.04% -> 17.34% | Improved |
| apply VTO on Pro-Pose avatar instead of original image | VTO · WPose user study | person error: 2.51% -> 3.02% | Hurt |
Closest papers by meaning
- Bringing Your Portrait to 3D Presence (CVPR 2026 · similarity 0.88)
- UP2You: Fast Reconstruction of Yourself from Unconstrained Photo Collections (ICLR 2026 · similarity 0.88)
- FRESA: Feedforward Reconstruction of Personalized Skinned Avatars from Few Images (CVPR 2025 · similarity 0.87)
- Visual Persona: Foundation Model for Full-Body Human Customization (CVPR 2025 · similarity 0.86)
- Vid2Avatar-Pro: Authentic Avatar from Videos in the Wild via Universal Prior (CVPR 2025 · similarity 0.86)
- PSHuman: Photorealistic Single-image 3D Human Reconstruction using Cross-Scale Multiview Diffusion and Explicit Remeshing (CVPR 2025 · similarity 0.85)
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.
