Research map / 3D vision / 3D generation
Learning Generatable Mutual Distance for Scene-Aware Human Motion Generation
Zonglin Yang, Chaoyue Xing, Yixuan Yin, Miaomiao Liu, Liyuan Pan
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 mutual distance representation (ablation) | MDNet · HUMANISE ablation | goal dist.: 0.298 -> 0.057 | Improved |
| add VLM semantic anchors (ablation) | MDNet · HUMANISE ablation | goal dist.: 0.365 -> 0.057 | Improved |
| add per-basis point distance (ablation) | MDNet · HUMANISE ablation | goal dist.: 0.158 -> 0.057 | Improved |
| add per-joint signed distance (ablation) | MDNet · HUMANISE ablation | penetration mean: 0.992 -> 0.635 | Improved |
| model mutual distance in DCT frequency domain (ablation) | MDNet · HUMANISE ablation | penetration mean: 0.741 -> 0.635 | Improved |
| use mutual distance instead of random orthogonal projection of it (ablation) | MDNet · HUMANISE ablation | goal dist.: 0.147 -> 0.057 | Improved |
| add semantic anchors to Wang et al. | Wang et al. · HUMANISE (GT cropping) | goal dist.: 0.156 -> 0.057 | Improved |
| use Gemini-1.5-Pro instead of other VLMs for anchors (ablation) | MDNet · HUMANISE anchor VLM | goal dist.: 0.390 (o4-mini) -> 0.057 | Improved |
| MDNet vs Wang et al. on contact | MDNet · HUMANISE test | contact: 95.86 -> 95.05 | Hurt |
| MDNet zero-shot transfer to PROX | MDNet · PROX (zero-shot) | goal dist.: 0.283 -> 0.054 vs Wang et al. | Improved |
| replace GT scene cropping with VLM-predicted cropping (ablation) | MDNet · HUMANISE | goal dist.: 0.055 -> 0.057 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Humanise — HUMANISE: Language-conditioned Human Motion Generation in 3D Scenes (NeurIPS 2022) · better on 5 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Move as You Say Interact as You Can: Language-guided Human Motion Generation with Scene Affordance (CVPR 2024 · similarity 0.91)
- HSI-GPT: A General-Purpose Large Scene-Motion-Language Model for Human Scene Interaction (CVPR 2025 · similarity 0.89)
- Generating Human Motion in 3D Scenes from Text Descriptions (CVPR 2024 · similarity 0.88)
- CG-HOI: Contact-Guided 3D Human-Object Interaction Generation (CVPR 2024 · similarity 0.88)
- Motion-Agent: A Conversational Framework for Human Motion Generation with LLMs (ICLR 2025 · similarity 0.87)
- Seamless Human Motion Composition with Blended Positional Encodings (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.
