Research map / Video / Video generation
FineMotion: A Dataset and Benchmark with both Spatial and Temporal Annotation for Fine-grained Motion Generation and Editing
Bizhu Wu, Jinheng Xie, Meidan Ding, Zhe Kong, Jianfeng Ren, Ruibin Bai, Rong Qu, Linlin Shen
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 fine-grained body-part text (DT) to training and input | T2M-GPT · FineMotion benchmark | FID: 0.123 -> 0.091 | Improved |
| add fine-grained body-part text (DT) to training and input | T2M-GPT · FineMotion benchmark | R-Precision Top-3: 0.781 -> 0.789 | Improved |
| add fine-grained body-part text (DT) to training and input | MDM · FineMotion benchmark | R-Precision Top-3: 0.606 -> 0.759 | Improved |
| add fine-grained body-part text (DT) to training and input | MDM · FineMotion benchmark | FID: 3.137 -> 0.488 | Improved |
| add fine-grained body-part text (DT) to training and input | MoMask · FineMotion benchmark | FID: 0.249 -> 0.055 | Improved |
| add fine-grained body-part text (DT) to training and input | MoMask · FineMotion benchmark | R-Precision Top-3: 0.753 -> 0.813 | Improved |
| encode coarse and detailed text separately (T&DT) rather than concatenated (TDT) (ablation) | MoMask (BPMSD) · FineMotion benchmark | FID: 8.328 -> 0.088 | Improved |
| encode coarse and detailed text separately (T&DT) rather than concatenated (TDT) (ablation) | MoMask (BPMP) · FineMotion benchmark | R-Precision Top-3: 0.628 -> 0.813 | Improved |
| use human-annotated DT in training instead of only automatic DT (ablation) | T2M-GPT (BPMSD) · FineMotion benchmark | FID: 0.091 -> 0.091 | No clear effect |
| add detailed-text multitask training, evaluated with coarse text only | T2M-GPT · HumanML3D-style T2M test | FID: 0.123 -> 0.154 | Hurt |
| add detailed-text multitask training, evaluated with coarse text only | MoMask · HumanML3D-style T2M test | FID: 0.249 -> 0.120 | Improved |
| add detailed-text multitask training, evaluated with coarse text only | MDM · HumanML3D-style T2M test | R-Precision Top-3: 0.606 -> 0.746 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- FineMoGen — FineMoGen: Fine-Grained Spatio-Temporal Motion Generation and Editing (NeurIPS 2023) · better on 2 of 2 shared results
- Fg-T2M — Fg-T2M: Fine-Grained Text-Driven Human Motion Generation via Diffusion Model (ICCV 2023) · better on 2 of 2 shared results
- TEMOS — TEMOS: Generating Diverse Human Motions from Textual Descriptions (ECCV 2022) · 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
- FrankenMotion: Part-level Human Motion Generation and Composition (CVPR 2026 · similarity 0.92)
- OpenT2M: No-frill Motion Generation with Open-source, Large-scale, High-quality Data (CVPR 2026 · similarity 0.91)
- Fg-T2M++: LLMs-Augmented Fine-Grained Text Driven Human Motion Generation (IJCV 2025 · similarity 0.91)
- HumanTOMATO: Text-aligned Whole-body Motion Generation (ICML 2024 · similarity 0.90)
- Scaling Large Motion Models with Million-Level Human Motions (ICML 2025 · similarity 0.89)
- SnapMoGen: Human Motion Generation from Expressive Texts (NeurIPS 2025 · similarity 0.89)
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.
