EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

MoR–Swin: Efficient Vision Transformer Using Mixture of Recursions

Yongbao Ai iD, Tianxiang Gao, Zhipeng Lin, Longqi Yang, Qingyu Chang

Journal of Imaging · 12.08.2026 · 12 (8) · 377

DOI: 10.3390/jimaging12080377 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Imaging, online erschienen am 12.08.2026. Verfasst von Yongbao Ai und 4 weiteren Autor:innen. Beteiligte Einrichtung: Academy of Military Science, Institute of Military Intelligence, Beijing 100091, China. Die Arbeit nennt 19 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Vision Transformers, especially Swin Transformer, have become default backbones for various vision tasks but suffer from high memory consumption and training costs. This letter proposes MoR–Swin, a novel architecture that integrates Mixture of Recursions (MoR) into Swin Transformer. An adaptive token-level recursion mechanism dynamically allocates computational depth based on semantic complexity. A recursive window attention module and a lightweight router with load balancing loss are introduced. Extensive experiments on ImageNet classification, COCO detection, and ADE20K segmentation show that MoR–Swin reduces parameters by about 50% and accelerates inference up to twofold at a modest accuracy cost (within about 0.5 points of Swin-B on ImageNet-1K). It provides a new technical pathway for optimizing Vision Transformer models, significantly enhancing their applicability in resource-constrained environments.

Schlagwörter

Stichwörter der Autor:innen

Swin Transformeradaptive computationmixture of recursionsparameter efficiency

Autor:innen und Institutionen

  • Academy of Military Science, Institute of Military Intelligence, Beijing 100091, China.

ORCID-Profile: Yongbao Ai

Förderung

  • National Natural Science Foundation of China

Zitieren

APA

Ai, Y., Gao, T., Lin, Z., Yang, L., & Chang, Q. (2026). MoR–Swin: Efficient Vision Transformer Using Mixture of Recursions. Journal of Imaging, 12(8), 377. https://doi.org/10.3390/jimaging12080377

Vancouver

Ai Y, Gao T, Lin Z, Yang L, Chang Q. MoR–Swin: Efficient Vision Transformer Using Mixture of Recursions. Journal of Imaging. 2026;12(8):377. doi:10.3390/jimaging12080377

BibTeX
@article{Ai2026,
  author = {Ai, Yongbao and Gao, Tianxiang and Lin, Zhipeng and Yang, Longqi and Chang, Qingyu},
  title = {MoR–Swin: Efficient Vision Transformer Using Mixture of Recursions},
  journal = {Journal of Imaging},
  year = {2026},
  volume = {12},
  number = {8},
  pages = {377},
  doi = {10.3390/jimaging12080377}
}
Kostenlos starten →