Research map / Video / Video generation
CoMA: Compositional Human Motion Generation with Multi-modal Agents
Shanlin Sun, Jiaqi Xu, Gabriel de Araujo, Shenghan Zhou, Hanwen Zhang, Ziheng Huang, Chenyu You, Xiaohui Xie
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 LLM Task Planner agent (recaption, segmentation, decomposition) on top of SPAM alone (ablation) | SPAM · MTT complex motion | R@1: 12.0 -> 30.1 | Improved |
| add Task Planner to SPAM (ablation) | SPAM · MTT complex motion | FID: 0.57 -> 0.48 | Improved |
| add Motion Reviewer self-correction (ablation) | SPAM + Task Planner · MTT complex motion | R@1: 30.1 -> 32.0 | Improved |
| add Motion Reviewer self-correction (ablation) | SPAM + Task Planner · MTT complex motion | FID: 0.48 -> 0.49 | Hurt |
| add further agent stage (+Task) to pipeline (ablation) | CoMA · MTT complex motion | R@3: 53.1 -> 54.6 | Improved |
| use spatially-aware SPAM instead of MoMask for generation | text-to-motion · HumanML3D test | R-Precision Top-1: 0.521 -> 0.526 | Improved |
| use SPAM vs MoMask on FID | text-to-motion · HumanML3D test | FID: 0.045 -> 0.092 | Hurt |
| use full CoMA pipeline instead of baselines | complex long prompts · User study, 42 prompts | user rating | Improved |
| use MVC motion captioner instead of MotionLLM | motion captioning · HumanML3D motion captioning | Rouge: 48.7 -> 45.79 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 3 on most shared results.
- MotionLLM — MotionLLM: Understanding Human Behaviors from Human Motions and Videos (TPAMI 2025) · better on 2 of 4 shared results
- Motion-R1 — Motion-R1: Enhancing Motion Generation with Decomposed Chain-of-Thought and RL Binding (ICLR 2026) · better on 0 of 2 shared results
- FineMoGen — FineMoGen: Fine-Grained Spatio-Temporal Motion Generation and Editing (NeurIPS 2023) · better on 1 of 1 shared result
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
- Motion-Agent: A Conversational Framework for Human Motion Generation with LLMs (ICLR 2025 · similarity 0.90)
- InterControl: Zero-shot Human Interaction Generation by Controlling Every Joint (NeurIPS 2024 · similarity 0.88)
- Seamless Human Motion Composition with Blended Positional Encodings (CVPR 2024 · similarity 0.87)
- MMM: Generative Masked Motion Model (CVPR 2024 · similarity 0.87)
- AMD: Autoregressive Motion Diffusion (AAAI 2024 · similarity 0.87)
- Everything2Motion: Synchronizing Diverse Inputs via a Unified Framework for Human Motion Synthesis (AAAI 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.
