ZeyuLing/Motius-TM2T-HumanML3D

TM2T Model Card

Tokenized reciprocal motion-language translation, reproduced as a standalone Motius M2T pipeline.

Paper | Project Page | Original GitHub | Motius Checkpoint

TM2T is the ECCV 2022 reciprocal text-to-motion and motion-to-text method. The Motius release contains the HumanML3D VQ tokenizer, motion-to-text Transformer, vocabulary, statistics, and inference runtime. It does not import an original repository checkout.

Release Snapshot

Item	Value
Released task	M2T
Motion representation	HumanML3D-263, 20 fps
Motion tokenizer	1,024-code VQ tokenizer
Caption model	4-layer encoder / 4-layer decoder Transformer
Decoding	Beam search, beam size 2
Checkpoint	ZeyuLing/Motius-TM2T-HumanML3D
Pipeline	motius.pipelines.tm2t.TM2TPipeline

Usage

import numpy as np
from motius.pipelines.tm2t import TM2TPipeline

pipe = TM2TPipeline.from_pretrained(
    "ZeyuLing/Motius-TM2T-HumanML3D",
    bundle_kwargs={"device": "cuda"},
)
motion = np.load("sample.npy")  # denormalized HumanML3D-263
caption = pipe.infer_m2t([motion], lengths=[len(motion)])[0]

M2T Evaluation

Full 4,400-sample evaluation follows the shared HumanML3D M2T protocol. Results are published only after the complete prediction set and metric artifact pass the population and sample-ID checks.

Samples	BLEU-4	ROUGE-L	CIDEr	BERT F1	R@1	R@2	R@3	Matching
4,400	-	-	-	-	-	-	-	-

Motion Representation

TM2T normalizes HumanML3D-263 features with its released training statistics. The VQ encoder removes four contact dimensions, maps each clip to discrete motion tokens, and the reciprocal Transformer translates those tokens to text.

Motius Components

Component	Path
Pipeline	motius/pipelines/tm2t/pipeline.py
Bundle	motius/models/tm2t/bundle.py
Runtime	motius/models/tm2t/network.py
License	motius/models/tm2t/LICENSE

Citation

@inproceedings{guo2022tm2t,
  title={TM2T: Stochastic and Tokenized Modeling for the Reciprocal Generation of 3D Human Motions and Texts},
  author={Guo, Chuan and Zuo, Xinxin and Wang, Sen and Cheng, Li},
  booktitle={European Conference on Computer Vision},
  year={2022}
}