Research map / 3D vision / Human pose estimation
H4D: Human 4D Modeling by Learning Neural Compositional Representation
Boyan Jiang, Yinda Zhang, Xingkui Wei, Xiangyang Xue, Yanwei Fu
Check 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 |
|---|---|---|---|
| Use H4D instead of 4D-CR | 4D human representation · CAPE 4D reconstruction | IoU: 62.9% -> 73.3% | Improved |
| Use H4D instead of 4D-CR | 4D human representation · CAPE motion retargeting | IoU: 47.3% -> 70.7% | Improved |
| Use H4D instead of 4D-CR | 4D human representation · CAPE motion completion | IoU: 76.6% -> 90.3% | Improved |
| Use H4D instead of 4D-CR | 4D human representation · CAPE future prediction | IoU: 64.0% -> 71.7% | Improved |
| Combine the linear motion prior with learned per-frame compensation (ablation) | H4D · CAPE motion recovery | motion accuracy | Improved |
| Add Motion-Comp to the linear motion model (ablation) | H4D · CAPE motion recovery | motion accuracy | Improved |
| Add auxiliary-code-driven Shape-Comp (ablation) | H4D · CAPE clothed mesh reconstruction | PVE | Improved |
| Use a GRU temporal encoder instead of modified PointNet (ablation) | H4D · CAPE reconstruction | reconstruction performance | Improved |
| Use GRU-based Shape-Comp instead of the naive CAPE decoder | H4D auto-decoding · CAPE motion completion with noisy observations | PVE | Improved |
| Keep VAE and adversarial loss in Shape-Comp (ablation) | H4D · CAPE shape compensation | reconstruction performance | Hurt |
| Use H4D instead of 4D-CR-SMPL for forward shape and motion recovery | human body estimation · CAPE shape and motion recovery (forward) | PA-MPJPE: 59.2 -> 38.4 | Improved |
| Use H4D instead of 4D-CR-SMPL for backward completion | human body estimation · CAPE motion completion (backward) | MPJPE: 67.3 -> 42.7 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- VIBE — VIBE: Video Inference for Human Body Pose and Shape Estimation (CVPR 2020) · better on 3 of 3 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
- H4MER: Human 4D Modeling by Learning Neural Compositional Representation With Transformer (TPAMI 2023 · similarity 0.96)
- Learning Compositional Representation for 4D Captures With Neural ODE (CVPR 2021 · similarity 0.87)
- LoRD: Local 4D Implicit Representation for High-Fidelity Dynamic Human Modeling (ECCV 2022 · similarity 0.87)
- H-NeRF: Neural Radiance Fields for Rendering and Temporal Reconstruction of Humans in Motion (NeurIPS 2021 · similarity 0.86)
- SurMo: Surface-based 4D Motion Modeling for Dynamic Human Rendering (CVPR 2024 · similarity 0.86)
- H-Flow: Self-supervised Human Scene Flow via Physics-inspired Joint Multi-modal Learning (NeurIPS 2026 · 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.
