Research map / Video / Video generation
Fg-T2M++: LLMs-Augmented Fine-Grained Text Driven Human Motion Generation
Yin Wang, Li Mu, Jiapeng Liu, Zhiying Leng, Frederick W. B. Li, Ziyao Zhang, Xiaohui Liang
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 word semantic parsing to text prompt (ablation) | Fg-T2M++ · KIT-ML test | FID: 0.344 -> 0.259 | Improved |
| add LLM action body parsing (ablation) | Fg-T2M++ · KIT-ML test | FID: 0.259 -> 0.135 | Improved |
| replace standard transformer with syntax-tree GCN (ablation) | Fg-T2M++ text encoder · KIT-ML test | FID: 0.724 -> 0.357 | Improved |
| replace GCN with hyperbolic GCN (ablation) | Fg-T2M++ text encoder · KIT-ML test | FID: 0.357 -> 0.164 | Improved |
| add cross-perception module (ablation) | Fg-T2M++ text encoder · KIT-ML test | FID: 0.164 -> 0.135 | Improved |
| add word-level LLM reference (ablation) | Fg-T2M++ fusion · KIT-ML test | FID: 0.281 -> 0.225 | Improved |
| add sentence-level feature fusion (ablation) | Fg-T2M++ fusion · KIT-ML test | FID: 0.225 -> 0.167 | Improved |
| add sentence-level reference (ablation) | Fg-T2M++ fusion · KIT-ML test | FID: 0.167 -> 0.135 | Improved |
| add all Fg-T2M++ components to Fg-T2M | Fg-T2M · KIT-ML test | R-Precision Top-1: 0.418 -> 0.442 | Improved |
| replace GPT-3.5 with GPT-4 as parser | Fg-T2M++ · KIT-ML test (LLM choice) | FID: 0.73 -> 0.68 | Improved |
| use Fg-T2M++ instead of ReMoDiffuse | Fg-T2M++ · HumanML3D test | FID: 0.103 -> 0.089 | Improved |
| use Fg-T2M++ instead of ReMoDiffuse | Fg-T2M++ · HumanML3D test | Diversity: 9.018 -> 9.223 | Hurt |
Later papers that compared with this method
2 later accepted papers report FG-T2M++ as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- Think Before You Move: Latent Motion Reasoning for Text-to-Motion Generation (TPAMI 2026 · better on 1 of 1 shared result)
- FineMotion: A Dataset and Benchmark with both Spatial and Temporal Annotation for Fine-grained Motion Generation and Editing (ICCV 2025 · better on 2 of 2 shared results)
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.
- FineMoGen — FineMoGen: Fine-Grained Spatio-Temporal Motion Generation and Editing (NeurIPS 2023) · better on 4 of 4 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
- Fg-T2M: Fine-Grained Text-Driven Human Motion Generation via Diffusion Model (ICCV 2023 · similarity 0.92)
- Open the Motion Door: Atomic Motion Decomposition and Recomposition for Open-Vocabulary Motion Generation (CVPR 2026 · similarity 0.91)
- Text Motion Translator: A Bi-Directional Model for Enhanced 3D Human Motion Generation from Open-Vocabulary Descriptions (ECCV 2024 · similarity 0.91)
- FineMotion: A Dataset and Benchmark with both Spatial and Temporal Annotation for Fine-grained Motion Generation and Editing (ICCV 2025 · similarity 0.91)
- HumanTOMATO: Text-aligned Whole-body Motion Generation (ICML 2024 · similarity 0.90)
- Shape My Moves: Text-Driven Shape-Aware Synthesis of Human Motions (CVPR 2025 · 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.
