Research map / Video / Video generation
RoMo: A Large-Scale, Richly Organized Dataset and Semantic Taxonomy for Human Motion Generation
Jiahao Zhang, Joseph Liu, Young-Yoon Lee, Seonghyeon Moon, Victor Zordan, Guy Tevet, C. Karen Liu, Stephen Gould, Oren Jacob, Haomiao Jiang, Mubbasir Kapadia, Yizhak Ben-Shabat
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 |
|---|---|---|---|
| train an autoregressive GPT model (MMGPT) instead of a diffusion model (MDM) on RoMo | RoMo text-to-motion · RoMo test (text-to-motion) | FID: 20.63 -> 12.80 | Improved |
| train MMGPT instead of MDM on RoMo | RoMo text-to-motion · RoMo test (text-to-motion) | Matching Score: 12.06 -> 22.08 | Improved |
| train MMGPT instead of MDM on RoMo | RoMo text-to-motion · RoMo test (text-to-motion) | Diversity: 27.67 -> 16.68 | Hurt |
| train MMGPT instead of MDM on RoMo | RoMo text-to-motion · RoMo test (text-to-motion) | Foot Skating (x1e-3): 1.70 -> 92.0 | Hurt |
| train MMGPT instead of MDM on RoMo | RoMo text-to-motion · RoMo test (text-to-motion) | Ground Penetration (x1e-5): 0.0 -> 3.55 | Hurt |
| train MMGPT instead of MDM on RoMo | RoMo text-to-motion · RoMo test (text-to-motion) | Dynamic Score: 0.2138 -> 0.3268 | Improved |
| train MMGPT instead of MDM on RoMo | RoMo text-to-motion · RoMo test (text-to-motion) | Floating (x1e-2): 1.67 -> 0.0311 | Improved |
| use taxonomy-aware adaptive filtering and curation (RoMo) instead of MotionMillion's minimal curation | motion dataset · mean dynamic score | mean dynamic score: 0.222 -> 0.336 | Improved |
| build dataset from taxonomy-driven queries (RoMo) vs MotionMillion | motion dataset · Taxonomy subcategory coverage | subcategories covered: 1,277 -> 2,065 (61.7% increase) | Improved |
Closest papers by meaning
- Scaling Large Motion Models with Million-Level Human Motions (ICML 2025 · similarity 0.90)
- The Quest for Generalizable Motion Generation: Data, Model, and Evaluation (ICLR 2026 · similarity 0.90)
- OpenT2M: No-frill Motion Generation with Open-source, Large-scale, High-quality Data (CVPR 2026 · similarity 0.90)
- AMD: Autoregressive Motion Diffusion (AAAI 2024 · similarity 0.89)
- How to Move Your Dragon: Text-to-Motion Synthesis for Large-Vocabulary Objects (ICML 2025 · similarity 0.88)
- SnapMoGen: Human Motion Generation from Expressive Texts (NeurIPS 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.
