Research map / 3D vision / 3D generation
Stochastic Scene-Aware Motion Prediction
Mohamed Hassan, Duygu Ceylan, Ruben Villegas, Jun Saito, Jimei Yang, Yi Zhou, Michael J. Black
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 explicit path planning module (ablation) | SAMP · Cluttered scene | penetration frames: 11.2% -> 3.8% | Improved |
| explicit path planning instead of volumetric sensor (NSM) (ablation) | SAMP · Cluttered scene | penetration frames: 8.11% -> 3.8% | Improved |
| add Interaction Encoder (ablation) | SAMP · Goal reach precision | position error: 14.82 cm -> 6.09 cm | Improved |
| add Interaction Encoder (ablation) | SAMP · Goal reach precision | rotation error: 3.65 deg -> 3.55 deg | Mixed |
| use SAMP MotionNet instead of MLP (ablation) | SAMP · Liedown goal reaching | success: inf -> 5.76 cm | Improved |
| use SAMP instead of MoE (ablation) | SAMP · Sit position error | PE (cm): 7.99 -> 6.09 | Improved |
| use SAMP instead of MLP (ablation) | SAMP · Liedown FD | FD: 36.85 -> 17.30 | Improved |
| use SAMP instead of MLP (ablation) | SAMP · Sit execution time | time (s): 13.06 -> 12.53 | Improved |
| use SAMP stochastic model instead of deterministic NSM (ablation) | SAMP · Sit diversity | diversity: 0.0 -> 0.44 | Improved |
| SAMP versus NSM without phase labels (ablation) | SAMP · Sit FD | FD: 4.21 -> 6.20 | Hurt |
| SAMP versus NSM on execution time (ablation) | SAMP · Sit | execution time (s): 10.26 -> 12.93 | Hurt |
| add schedule sampling in training (ablation) | SAMP · goal reaching | reaching goal | Improved |
Closest papers by meaning
- Harmonizing Stochasticity and Determinism: Scene-responsive Diverse Human Motion Prediction (NeurIPS 2024 · similarity 0.89)
- CoMusion: Towards Consistent Stochastic Human Motion Prediction via Motion Diffusion (ECCV 2024 · similarity 0.89)
- Towards Diverse and Natural Scene-Aware 3D Human Motion Synthesis (CVPR 2022 · similarity 0.88)
- BeLFusion: Latent Diffusion for Behavior-Driven Human Motion Prediction (ICCV 2023 · similarity 0.88)
- Learning to Sit: Synthesizing Human-Chair Interactions via Hierarchical Control (AAAI 2021 · similarity 0.87)
- MotionMap: Representing Multimodality in Human Pose Forecasting (CVPR 2025 · 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.
