Research map / 3D vision / Human pose estimation
Human Motion Prediction via Learning Local Structure Representations and Temporal Dependencies
Read the paperCheck 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 five component-specific branches for body parts (ablation) | SkelNet · Human3.6M long-term | MoF: 1.23 -> 1.17 | Helped |
| Use five component-specific branches for body parts (ablation) | SkelNet · CMU mocap long-term | MoF: 0.94 -> 0.89 | Helped |
| Keep the residual connection (ablation) | SkelNet · Human3.6M long-term | MoF: 2.48 -> 1.17 | Helped |
| Keep the residual connection (ablation) | SkelNet · CMU mocap long-term | MoF: 1.77 -> 0.89 | Helped |
| Use five separate body-part groups rather than three-group UD or LR layouts (ablation) | SkelNet · Human3.6M long-term | MoF: 1.19 -> 1.17; 1.21 -> 1.17 | Helped |
| Add long prior information (ablation) | SkelNet · Human3.6M long-term | MoF: 1.17 -> 1.19 | Hurt |
| Add long prior information (ablation) | SkelNet · CMU mocap long-term | MoF: 0.89 -> 0.87 | Helped |
| Use converging loss instead of sampling-based loss (ablation) | C-RNN · Human3.6M short-term | MoF: 0.79 -> 0.76 | Helped |
| Use converging loss instead of sampling-based loss (ablation) | C-RNN · CMU mocap short-term | MoF: 0.63 -> 0.59 | Helped |
| Combine the spatial and temporal models with the trained Merging Network (ablation) | Skel-TNet · Human3.6M short-term | MoF: 0.78 -> 0.73; 0.79 -> 0.73; 0.80 -> 0.73 | Helped |
| Combine the spatial and temporal models with the trained Merging Network (ablation) | Skel-TNet · CMU mocap short-term | MoF: 0.60 -> 0.55; 0.59 -> 0.55; 0.62 -> 0.55 | Helped |
| Train the three components in two joint arrangements instead of separately (ablation) | Skel-TNet · Human3.6M short-term | MoF: 0.82; 0.83 vs 0.73 | Hurt |
Closest papers by meaning
- Deep Representation Learning for Human Motion Prediction and Classification (CVPR 2017 · similarity 0.89)
- Context-Aware Human Motion Prediction (CVPR 2020 · similarity 0.88)
- Learning Dynamic Relationships for 3D Human Motion Prediction (CVPR 2020 · similarity 0.88)
- Progressively Generating Better Initial Guesses Towards Next Stages for High-Quality Human Motion Prediction (CVPR 2022 · similarity 0.87)
- Revealing Key Details to See Differences: A Novel Prototypical Perspective for Skeleton-based Action Recognition (CVPR 2025 · similarity 0.86)
- ALIEN: Implicit Neural Representations for Human Motion Prediction under Arbitrary Latency (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.