Research map / 3D vision / 3D generation
Unleashing Guidance Without Classifiers for Human-Object Interaction Animation
Ziyin Wang, Sirui Xu, chuan guo, Bing Zhou, Jiangshan Gong, Jian Wang, Yu-Xiong Wang, Liangyan Gui
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 pace-induced guidance at inference (ablation) | LIGHT · InterAct | FID: 0.196 -> 0.148 | Improved |
| add pace-induced guidance (ablation) | LIGHT · InterAct | R-Precision Top-1: 0.395 -> 0.421 | Improved |
| add pace-induced guidance (ablation) | LIGHT · InterAct | Contact F1: 0.599 -> 0.627 | Improved |
| add pace-induced guidance (ablation) | LIGHT · InterAct | Pene: 0.121 -> 0.132 | Mixed |
| add pace-induced guidance (ablation) | LIGHT · InterAct | FSR: 0.066 -> 0.078 | Hurt |
| separate hand and body tokens (ablation) | LIGHT · InterAct | Contact F1: 0.611 -> 0.627 | Improved |
| separate human and object tokens (ablation) | LIGHT · InterAct | Contact F1: 0.572 -> 0.627 | Improved |
| separate human and object tokens (ablation) | LIGHT · InterAct | Pene: 0.105 -> 0.132 | Hurt |
| add object-level data augmentation (ablation) | LIGHT · unseen in-category objects | R-Precision Top-1: 0.216 -> 0.279 | Improved |
| add object-level data augmentation (ablation) | LIGHT · unseen cross-category objects | Contact F1: 0.351 -> 0.560 | Improved |
| add object-level data augmentation (ablation) | LIGHT · unseen cross-category objects | R-Precision Top-1: 0.022 -> 0.022 | No clear effect |
| LIGHT guidance vs text CFG alignment with depenetration gradient | LIGHT · InterAct | cosine similarity: 0.217 -> 0.217 + 0.035 | Improved |
Closest papers by meaning
- CG-HOI: Contact-Guided 3D Human-Object Interaction Generation (CVPR 2024 · similarity 0.90)
- HOIAnimator: Generating Text-prompt Human-object Animations using Novel Perceptive Diffusion Models (CVPR 2024 · similarity 0.90)
- CoDA: Coordinated Diffusion Noise Optimization for Whole-Body Manipulation of Articulated Objects (NeurIPS 2025 · similarity 0.88)
- NIFTY: Neural Object Interaction Fields for Guided Human Motion Synthesis (CVPR 2024 · similarity 0.88)
- InterDreamer: Zero-Shot Text to 3D Dynamic Human-Object Interaction (NeurIPS 2024 · similarity 0.88)
- InterAct: Advancing Large-Scale Versatile 3D Human-Object Interaction Generation (CVPR 2025 · similarity 0.88)
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.
