Learning Joint Representation of Human Motion and Language

Contrastive learning for shared motion–language representations.

MoLang learns a shared representation of human motion and language from paired and unpaired data. The contrastive motion–language model supports action recognition and motion retrieval within a single representation space. (Kim et al., 2022)

Resources

PDF · arXiv

References

2022

arXiv

Learning Joint Representation of Human Motion and Language

Jihoon Kim, Youngjae Yu, Seungyoun Shin, and 2 more authors

arXiv preprint arXiv:2210.15187, Oct 2022

DOI arXiv Bib PDF

@article{kim2022learning,
  title = {Learning Joint Representation of Human Motion and Language},
  author = {Kim, Jihoon and Yu, Youngjae and Shin, Seungyoun and Byun, Taehyun and Choi, Sungjoon},
  journal = {arXiv preprint arXiv:2210.15187},
  month = oct,
  year = {2022},
  doi = {10.48550/arXiv.2210.15187},
}