EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Latent space improved masked reconstruction model for human skeleton-based action recognition

Enqing Chen, Xueting Wang, Xin Guo, Ying Zhu, Dong Li

Frontiers in Neurorobotics · 12.02.2025 · 19

DOI: 10.3389/fnbot.2025.1482281 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Frontiers in Neurorobotics, online erschienen am 12.02.2025. Verfasst von Enqing Chen und 4 weiteren Autor:innen. Beteiligte Einrichtung: School of Electrical and Information Engineering, Zhengzhou University, Zhengzhou, China. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 43 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Human skeleton-based action recognition is an important task in the field of computer vision. In recent years, masked autoencoder (MAE) has been used in various fields due to its powerful self-supervised learning ability and has achieved good results in masked data reconstruction tasks. However, in visual classification tasks such as action recognition, the limited ability of the encoder to learn features in the autoencoder structure results in poor classification performance. We propose to enhance the encoder's feature extraction ability in classification tasks by leveraging the latent space of variational autoencoder (VAE) and further replace it with the latent space of vector quantized variational autoencoder (VQVAE). The constructed models are called SkeletonMVAE and SkeletonMVQVAE, respectively. In SkeletonMVAE, we constrain the latent variables to represent features in the form of distributions. In SkeletonMVQVAE, we discretize the latent variables. These help the encoder learn deeper data structures and more discriminative and generalized feature representations. The experiment results on the NTU-60 and NTU-120 datasets demonstrate that our proposed method can effectively improve the classification accuracy of the encoder in classification tasks and its generalization ability in the case of few labeled data. SkeletonMVAE exhibits stronger classification ability, while SkeletonMVQVAE exhibits stronger generalization in situations with fewer labeled data.

Schlagwörter

Stichwörter der Autor:innen

human skeleton-based action recognitionmasked reconstruction modelself-supervised learningvariational autoencodervector quantized variational autoencoder

Autor:innen und Institutionen

  • School of Electrical and Information Engineering, Zhengzhou University, Zhengzhou, China.
  • State Grid Henan Electric Power Company Information and Communication Branch, Zhengzhou, China.

Zitieren

APA

Chen, E., Wang, X., Guo, X., Zhu, Y., & Li, D. (2025). Latent space improved masked reconstruction model for human skeleton-based action recognition. Frontiers in Neurorobotics, 19. https://doi.org/10.3389/fnbot.2025.1482281

Vancouver

Chen E, Wang X, Guo X, Zhu Y, Li D. Latent space improved masked reconstruction model for human skeleton-based action recognition. Frontiers in Neurorobotics. 2025;19. doi:10.3389/fnbot.2025.1482281

BibTeX
@article{Chen2025,
  author = {Chen, Enqing and Wang, Xueting and Guo, Xin and Zhu, Ying and Li, Dong},
  title = {Latent space improved masked reconstruction model for human skeleton-based action recognition},
  journal = {Frontiers in Neurorobotics},
  year = {2025},
  volume = {19},
  doi = {10.3389/fnbot.2025.1482281}
}
Kostenlos starten →