Research map / 3D vision / 3D generation
Revisit Human-Scene Interaction via Space Occupancy
Xinpeng Liu, Haowen Hou, Yanchao Yang, Yong-Lu Li, Cewu Lu
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 canonical occupancy input to the controller (ablation) | motion controller · MOB goal reaching | success rate (%): 71.22 -> 77.91 | Improved |
| add canonical occupancy input to the controller (ablation) | motion controller · MOB goal reaching | penetration (voxels per frame): 46.05 -> 15.21 | Improved |
| add penetration loss (ablation) | motion controller · MOB goal reaching | success rate (%): 74.14 -> 77.91 | Improved |
| add occupancy field loss (ablation) | motion controller · MOB goal reaching | foot sliding (%): 11.65 -> 8.13 | Improved |
| add field regulation module (ablation) | motion controller · MOB goal reaching | penetration (voxels per frame): 33.49 -> 15.21 | Improved |
| add field regulation module (ablation) | motion controller · MOB goal reaching | foot sliding (%): 6.57 -> 8.13 | Hurt |
| use voxel occupancy instead of BPS representation (ablation) | motion controller · MOB goal reaching | success rate (%): 73.64 -> 77.91 | Improved |
| add MOB pretraining data to CIRCLE training (ablation) | motion controller · CIRCLE test | success rate (%): 75.52 -> 78.97 | Improved |
| train on MOB only (no real scene data) versus CIRCLE only (ablation) | motion controller · CIRCLE test | success rate (%): 73.01 vs 75.52 | Mixed |
| use unified occupancy controller instead of NavMesh locomotion and interaction decomposition | DIMOS baseline · MOB goal reaching | success rate (%): 15.73 -> 77.91 | Improved |
| fine-tune controller on OMOMO versus OMOMO diffusion model | human-object interaction · OMOMO | Hand JPE: 24.01 -> 33.27 | Hurt |
Closest papers by meaning
- InterAct: Advancing Large-Scale Versatile 3D Human-Object Interaction Generation (CVPR 2025 · similarity 0.89)
- Scaling Up Dynamic Human-Scene Interaction Modeling (CVPR 2024 · similarity 0.89)
- HOI-M^3: Capture Multiple Humans and Objects Interaction within Contextual Environment (CVPR 2024 · similarity 0.87)
- InterDreamer: Zero-Shot Text to 3D Dynamic Human-Object Interaction (NeurIPS 2024 · similarity 0.86)
- GenZI: Zero-Shot 3D Human-Scene Interaction Generation (CVPR 2024 · similarity 0.86)
- CG-HOI: Contact-Guided 3D Human-Object Interaction Generation (CVPR 2024 · 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.
