Research map / 3D vision / 3D generation
SceneMI: Motion In-betweening for Modeling Human-Scene Interaction
Inwoo Hwang, Bing Zhou, Young Min Kim, Jian Wang, Chuan Guo
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 scene-awareness (global and local scene features) (ablation) | SceneMI · TRUMANS noise-free | MJPE All (m): 0.059 -> 0.023 | Improved |
| add scene-awareness (ablation) | SceneMI · TRUMANS close-proximity frames | collision frame ratio: 0.237 -> 0.162 | Improved |
| add global scene feature c_g (ablation) | SceneMI · TRUMANS noise-free | FID: 0.138 -> 0.123 | Improved |
| add local scene feature c_l (ablation) | SceneMI · TRUMANS noise-free | FID: 0.125 -> 0.123 | Improved |
| add noise-awareness (T*=20) (ablation) | SceneMI · TRUMANS synthetic noisy keyframes | FID: 0.157 -> 0.118 | Improved |
| increase T* beyond 20 (ablation) | SceneMI · TRUMANS synthetic noisy keyframes | MJPE All (m): 0.012 -> 0.015 | Hurt |
| add noise-awareness (ablation) | SceneMI · GIMO real-world | foot skating: 0.391 -> 0.163 | Improved |
| add scene-awareness (ablation) | SceneMI · GIMO real-world | collision frame ratio: 0.082 -> 0.060 | Improved |
| apply SceneMI in-betweening to raw GIMO motion | real-world GIMO data · GIMO real-world | foot skating: 0.261 -> 0.163 | Improved |
| add body shape conditioning (ablation) | SceneMI · TRUMANS noise-free (r=60) | MJPE All (m): 0.038 -> 0.023 | Improved |
| use larger scene grid 96x48x96 (ablation) | SceneMI · TRUMANS noise-free (r=60) | FID: 0.123 -> 0.130 | Hurt |
| vary keyframe interval from r=1 to r=60 (ablation) | SceneMI · TRUMANS noisy | FID: 0.122 -> 0.123 | No clear effect |
Closest papers by meaning
- HSI-GPT: A General-Purpose Large Scene-Motion-Language Model for Human Scene Interaction (CVPR 2025 · similarity 0.88)
- Scaling Up Dynamic Human-Scene Interaction Modeling (CVPR 2024 · similarity 0.87)
- SViMo: Synchronized Diffusion for Video and Motion Generation in Hand-object Interaction Scenarios (NeurIPS 2025 · similarity 0.87)
- InterAct: Advancing Large-Scale Versatile 3D Human-Object Interaction Generation (CVPR 2025 · similarity 0.87)
- InterDreamer: Zero-Shot Text to 3D Dynamic Human-Object Interaction (NeurIPS 2024 · similarity 0.86)
- HSI-GPT2: A Dual-Granularity Large Motion Reasoning Model with Diffusion Refinement for Human-Scene Interaction (CVPR 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.
