Research map / Video / Video generation
Priority-Centric Human Motion Generation in Discrete Latent Space
Hanyang Kong, Kehong Gong, Dongze Lian, Michael Bi Mi, Xinchao Wang
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 |
|---|---|---|---|
| replace linear noise schedule with dynamic-assessment priority-centric schedule (ablation) | M2DM · HumanML3D test | FID: 0.417 -> 0.352 | Improved |
| replace linear noise schedule with static-assessment priority-centric schedule (ablation) | M2DM · HumanML3D test | FID: 0.417 -> 0.395 | Improved |
| replace linear noise schedule with priority-centric schedule (ablation) | M2DM · KIT-ML test | FID: 0.502 -> 0.515 | No clear effect |
| add l2 normalization and orthogonal regularization to the VQ-VAE codebook (ablation) | Transformer VQ-VAE · HumanML3D test (reconstruction) | FID: 0.079 -> 0.063 | Improved |
| add orthogonal regularization on codebook entries | Transformer VQ-VAE · HumanML3D test | codebook usage | Improved |
| use priority-centric discrete diffusion for long text descriptions (>30 words) | text-to-motion · HumanML3D test, text >30 words | FID: 0.947 -> 0.356 | Improved |
| longer textual descriptions | MLD · HumanML3D test | R-Precision Top-1: 0.492 -> 0.289 | Hurt |
| use Transformer-based VQ-VAE with discrete diffusion instead of competing methods | M2DM · HumanML3D test | FID: 0.141 (T2M-GPT) vs 0.352 | Mixed |
Closest papers by meaning
- Rethinking Diffusion for Text-Driven Human Motion Generation: Redundant Representations, Evaluation, and Masked Autoregression (CVPR 2025 · similarity 0.93)
- AMD: Autoregressive Motion Diffusion (AAAI 2024 · similarity 0.92)
- Towards Detailed Text-to-Motion Synthesis via Basic-to-Advanced Hierarchical Diffusion Model (AAAI 2024 · similarity 0.92)
- Human Motion Diffusion as a Generative Prior (ICLR 2024 · similarity 0.90)
- Enhanced Fine-Grained Motion Diffusion for Text-Driven Human Motion Synthesis (AAAI 2024 · similarity 0.90)
- AMD: Anatomical Motion Diffusion with Interpretable Motion Decomposition and Fusion (AAAI 2024 · similarity 0.90)
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.
