EUVIMEDEUVIMED entdecken
Crossref

DSwinIR: Rethinking Window-Based Attention for Image Restoration

Gang Wu, Junjun Jiang, Kui Jiang, Xianming Liu, Liqiang Nie

IEEE Transactions on Pattern Analysis and Machine Intelligence · 2026-04 · 48 (4) · 4350-4366

DOI: 10.1109/tpami.2025.3646016 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Pattern Analysis and Machine Intelligence, 2026-04. Verfasst von Gang Wu und 4 weiteren Autor:innen. Die Arbeit wurde bisher 20-mal zitiert (Crossref) und nennt 140 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Natural Science Foundation of Heilongjiang Province of China for Excellent Youth Project.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Image restoration has witnessed significant advancements with the development of deep learning models. Transformer-based models, particularly those using window-based self-attention, have become a dominant force. However, their performance is constrained by the rigid, non-overlapping window partitioning scheme, which leads to insufficient feature interaction across windows and limited receptive fields. This highlights the need for more adaptive and flexible attention mechanisms. In this paper, we propose the Deformable Sliding Window Transformer for Image Restoration (DSwinIR), a new attention mechanism: the Deformable Sliding Window (DSwin) Attention. This mechanism introduces a token-centric and content-aware paradigm that moves beyond the grid and fixed window partition. It comprises two complementary components. First, it replaces the rigid partitioning with a token-centric sliding window paradigm, making it effective at eliminating boundary artifacts. Second, it incorporates a content-aware deformable sampling strategy, which allows the attention mechanism to learn data-dependent offsets and actively shape its receptive field to focus on the most informative image regions. Extensive experiments show that DSwinIR achieves strong results, including state-of-the-art performance on several evaluated benchmarks. For instance, in all-in-one image restoration, our DSwinIR surpasses the most recent backbone GridFormer by 0.53 dB on the three-task benchmark and 0.87 dB on the five-task benchmark.

Förderung

  • National Natural Science Foundation of China
  • Natural Science Foundation of Heilongjiang Province of China for Excellent Youth Project

Zitieren

APA

Wu, G., Jiang, J., Jiang, K., Liu, X., & Nie, L. (2026). DSwinIR: Rethinking Window-Based Attention for Image Restoration. IEEE Transactions on Pattern Analysis and Machine Intelligence, 48(4), 4350-4366. https://doi.org/10.1109/tpami.2025.3646016

Vancouver

Wu G, Jiang J, Jiang K, Liu X, Nie L. DSwinIR: Rethinking Window-Based Attention for Image Restoration. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2026;48(4):4350-4366. doi:10.1109/tpami.2025.3646016

BibTeX
@article{Wu2026,
  author = {Wu, Gang and Jiang, Junjun and Jiang, Kui and Liu, Xianming and Nie, Liqiang},
  title = {DSwinIR: Rethinking Window-Based Attention for Image Restoration},
  journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
  year = {2026},
  volume = {48},
  number = {4},
  pages = {4350--4366},
  doi = {10.1109/tpami.2025.3646016}
}
Kostenlos starten →