Research map / 3D vision / Face and shape
Neural Capture of Animatable 3D Human from Monocular Video
Gusi Te, Xiu Li, Xiao Li, Jinglu Wang, Wei Hu, Yan Lu
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 observation-space distance embedding (vs none) (ablation) | mesh-guided NeRF · People-Snapshot ablation (novel poses) | novel-pose PSNR: 27.51 -> 34.75 | Improved |
| use observation-space distance instead of canonical (rest-pose) distance (ablation) | mesh-guided NeRF · People-Snapshot ablation (novel poses) | novel-pose PSNR: 27.48 -> 34.75 | Improved |
| add relative direction embedding (ablation) | mesh-guided NeRF · People-Snapshot ablation (novel poses) | novel-pose PSNR: 28.15 -> 34.75 | Improved |
| inverse-transform the direction into rest space rather than keeping it in observation space (ablation) | mesh-guided NeRF · People-Snapshot ablation (novel poses) | novel-pose PSNR: 28.12 -> 34.75 | Improved |
| use geodesic 1-hop neighbors instead of Euclidean K-NN (ablation) | mesh-guided NeRF · People-Snapshot ablation (novel poses) | novel-pose PSNR: 30.69 -> 34.75 | Improved |
| use 1-hop neighborhood instead of 2-hop (ablation) | mesh-guided NeRF · People-Snapshot ablation (novel poses) | novel-pose PSNR: 31.97 -> 34.75 | Improved |
| use 1-hop neighborhood instead of only the nearest vertex (ablation) | mesh-guided NeRF · People-Snapshot ablation (novel poses) | novel-pose PSNR: 30.07 -> 34.75 | Improved |
| add joint pose refinement (ablation) | mesh-guided NeRF · Human3.6M | PSNR: 29.04 -> 31.51 | Improved |
| add joint pose refinement (ablation) | mesh-guided NeRF · People-Snapshot | PSNR: 32.99 -> 34.75 | Improved |
| add joint pose refinement (ablation) | mesh-guided NeRF · Human3.6M | SSIM: 0.951 -> 0.978 | Improved |
| add joint pose refinement (ablation) | mesh-guided NeRF · People-Snapshot | SSIM: 0.969 -> 0.972 | Improved |
| switch from AniNeRF to the proposed method (single view) | AniNeRF · ZJU-MoCap (single view) | PSNR: 23.75 -> 25.01 | Improved |
Closest papers by meaning
- MonoHuman: Animatable Human Neural Field From Monocular Video (CVPR 2023 · similarity 0.92)
- MPS-NeRF: Generalizable 3D Human Rendering From Multiview Images (TPAMI 2022 · similarity 0.91)
- H-NeRF: Neural Radiance Fields for Rendering and Temporal Reconstruction of Humans in Motion (NeurIPS 2021 · similarity 0.90)
- Neural Reconstruction of Relightable Human Model from Monocular Video (ICCV 2023 · similarity 0.90)
- Relightable and Animatable Neural Avatar from Sparse-View Video (CVPR 2024 · similarity 0.89)
- Joint Optimization of Neural Radiance Fields and Continuous Camera Motion from a Monocular Video (CVPR 2025 · similarity 0.88)
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.
