Research map / 3D vision / 3D generation
Stochastic Multi-Person 3D Motion Forecasting
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 |
|---|---|---|---|
| Add dual-level modeling (local/global separation) with discrete and continuous intent codes to CGAN | RNN SC-MPF, Transformer SC-MPF, Transformer MRT, Transformer XIA · CMU-Mocap 3-person and 2-person | FDE and ADE at 3s: e.g., MRT FDE 1.968 -> 1.823 at 3s | Helped |
| Remove separation of local and global levels (w/o Separation) versus full dual-level (ablation) | CGAN+DuMMF with MRT · CMU-Mocap POSE @3s | FDE: 0.448 vs 0.390 | Hurt |
| Remove discrete intent codes (ablation) | CGAN+DuMMF with MRT · CMU-Mocap POSE @3s | FDE: 0.490 vs 0.390 | Hurt |
| Remove continuous Gaussian intent codes (ablation) | CGAN+DuMMF with MRT · CMU-Mocap POSE @3s | FDE: 0.439 vs 0.390 | Hurt |
| Remove global social-level modeling (w/o Social) (ablation) | CGAN+DuMMF with MRT · CMU-Mocap ROOT @3s | FPD (diversity): FPD 1.661 vs 0.249 full; but ADE/FDE worse on POSE | Helped |
| Add DuMMF dual-level framework to DDPM | DDPM · SMPL-X AMASS, 75 frames | FDE: 11.111 -> 4.880 with 3 intents | Helped |
| Increase number of intents from 1 to 3 in DuMMF DDPM (ablation) | DuMMF DDPM · SMPL-X AMASS, 25 frames | ADE: 2.405 -> 1.456 | Helped |
| Dual-level DuMMF with a single intent vs pure DDPM | DDPM · SMPL-X AMASS, 75 frames | FDE: 11.111 -> 8.610 | Helped |
| Dual-level DuMMF with a single intent vs pure DDPM at short horizon (ablation) | DDPM · SMPL-X AMASS, 25 frames | ADE: 2.021 -> 2.405 (1 intent) | Hurt |
| Remove discrete intents then compare with Full on MuPoTS-3D generalization (ablation) | CGAN+DuMMF with MRT · MuPoTS-3D POSE @3s | FDE: 0.331 w/o Individual; 0.411 w/o Discrete vs 0.305 full | Hurt |
Closest papers by meaning
- Motion Diversification Networks (CVPR 2024 · similarity 0.91)
- FutureHuman3D: Forecasting Complex Long-Term 3D Human Behavior from Video Observations (CVPR 2024 · similarity 0.89)
- SimpliHuMoN: Simplifying Human Motion Prediction (NeurIPS 2026 · similarity 0.88)
- Harmonizing Stochasticity and Determinism: Scene-responsive Diverse Human Motion Prediction (NeurIPS 2024 · similarity 0.87)
- AMD: Autoregressive Motion Diffusion (AAAI 2024 · similarity 0.87)
- Multi-agent Long-term 3D Human Pose Forecasting via Interaction-aware Trajectory Conditioning (CVPR 2024 · similarity 0.87)
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.