Motion Captioning 벤치마크
Motion Captioning on HumanML3D
BLEU-4
- 2022-07-04 — TM2T: BLEU-4 22.3
- 2023-10-11 — ST-MLP: BLEU-4 25.0
| Rank | Model | BLEU-4 | BERTScore | Paper | Code | Year |
|---|---|---|---|---|---|---|
| 1 | ST-MLP | 25.0 | 40.3 | Guided Attention for Interpretable Motion Captioning | rd20karim/m2t-interpretable | 2023 |
| 2 | MLP+GRU | 23.4 | 37.2 | Motion2Language, unsupervised learning of synchronized semantic motion segmentation | rd20karim/M2T-Segmentation | 2023 |
| 3 | TM2T | 22.3 | 37.8 | TM2T: Stochastic and Tokenized Modeling for the Reciprocal Generation of 3D Human Motions and Texts | EricGuo5513/TM2T | 2022 |
| 4 | MotionGPT | 12.47 | 32.4 | MotionGPT: Human Motion as a Foreign Language | openmotionlab/motiongpt · openmeshlab/meshxl · open3da/ll3da | 2023 |