Research map / Video / Video generation
Towards Decompositional Human Motion Generation with Energy-Based Diffusion Models
Jianrong Zhang, Hehe Fan, Yi Yang
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 |
|---|---|---|---|
| finetune with DecompML-generated motions added (ablation) | SALAD · HumanML3D test | FID: 0.076 -> 0.060 | Improved |
| finetune with DecompML-generated motions added (ablation) | SALAD · HumanML3D test | Top-1: 0.581 -> 0.580 | No clear effect |
| finetune with DecompML-generated motions added (ablation) | EnergyMoGen · HumanML3D test | FID: 0.188 -> 0.147 | Improved |
| orthogonal self-supervised decomposition (OSS) training | SALAD-style latent diffusion · HumanML3D test | R-Precision Top-1: 0.581 -> 0.588 | Improved |
| compositional training (DEMOGEN-OSS latent) | SALAD · HumanML3D test | FID: 0.076 -> 0.092 | No clear effect |
| semantic-aware vs latent-aware EBM | DEMOGEN · HumanML3D test | FID: EXP: 0.078 vs 0.116; OSS: 0.092 vs 0.104 | Mixed |
| DEMOGEN-EXP compositional training | EnergyMoGen+SEF · MTT compositional | R@3: 28.0 -> 31.9 | Improved |
| DEMOGEN-EXP compositional training | EnergyMoGen+SEF · MTT compositional | FID: 0.604 -> 0.621 | Hurt |
| DEMOGEN-EXP compositional training | SALAD/EnergyMoGen · DecompML composition | Top-1: 0.342 -> 0.559 | Improved |
| DEMOGEN-OSS multi-concept training | EMG+AGD · MTT multi-concept | FID: 0.587 -> 0.580 | Improved |
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.
- EnergyMoGen — EnergyMoGen: Compositional Human Motion Generation with Energy-Based Diffusion Model in Latent Space (CVPR 2025) · better on 2 of 2 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
- Human Motion Diffusion as a Generative Prior (ICLR 2024 · similarity 0.90)
- MotionMix: Weakly-Supervised Diffusion for Controllable Motion Generation (AAAI 2024 · similarity 0.89)
- AMD: Autoregressive Motion Diffusion (AAAI 2024 · similarity 0.88)
- Rethinking Diffusion for Text-Driven Human Motion Generation: Redundant Representations, Evaluation, and Masked Autoregression (CVPR 2025 · similarity 0.88)
- AMD: Anatomical Motion Diffusion with Interpretable Motion Decomposition and Fusion (AAAI 2024 · similarity 0.88)
- MixerMDM: Learnable Composition of Human Motion Diffusion Models (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.
