Research map / 3D vision / Human pose estimation
Learning Dynamic Relationships for 3D Human Motion Prediction
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 dynamic natural-connectivity and global graphs instead of the fully connected FC-GCN graph | 3D human motion prediction · H3.6M Walking, 1000 ms | angle error: 0.79 -> 0.71 | Helped |
| Use dynamic natural-connectivity and global graphs instead of FC-GCN | 3D human motion prediction · 3DPW MoCap testing set | mean angle error: 0.64 -> 0.57 | Helped |
| Train and test directly on 3D coordinates rather than converting predicted angles | Ours · H3.6M Walking, 400 ms | MPJPE: 32.2 -> 29.9 | Helped |
| Remove adversarial learning (ablation) | Ours · H3.6M | mean angle error: 0.27 -> 0.24 | Hurt |
| Add a residual connection (ablation) | Ours · H3.6M | mean angle error: 0.35 -> 0.24 | Helped |
| Remove the global graph Q (ablation) | Ours · H3.6M | mean angle error: 0.24 -> 0.31 | Hurt |
| Remove the connective graph Ap (ablation) | Ours · H3.6M | mean angle error: 0.24 -> 0.28 | Hurt |
| Use both the connective graph Ap and global graph Q (ablation) | Ours · H3.6M | mean angle error: 0.31 -> 0.24 | Helped |
| Use a discriminator on the concatenated history and future instead of future-only sequences (ablation) | Ours · H3.6M | mean angle error: 0.50 -> 0.24 | Helped |
| Use a discriminator on the concatenated history and future (ablation) | Ours · H3.6M | mean 3D error: 89.3 -> 81.3 | Helped |
| Use the proposed model rather than FC-GCN | 3D human motion prediction · CMU MoCap Running, 400 ms | angle error: 0.74 -> 0.68 | Hurt |
Closest papers by meaning
- Deep Representation Learning for Human Motion Prediction and Classification (CVPR 2017 · similarity 0.90)
- We Are More Than Our Joints: Predicting How 3D Bodies Move (CVPR 2021 · similarity 0.89)
- Towards Natural and Accurate Future Motion Prediction of Humans and Animals (CVPR 2019 · similarity 0.89)
- FutureHuman3D: Forecasting Complex Long-Term 3D Human Behavior from Video Observations (CVPR 2024 · similarity 0.88)
- SAMoR: Motion Modelling for Articulated Objects of Any Skeleton and Topology (NeurIPS 2026 · similarity 0.88)
- Forecasting of 3D Whole-body Human Poses with Grasping Objects (CVPR 2024 · 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.