EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Remote Sensing Target Detection Network Based on Frequency Domain Enhanced Attention Mechanism With Mamba

Sibei Chen, Zhiwei Ye, Yiming Yang, Zijian He

IEEE Access · 2026 · 14 · 117453-117469

DOI: 10.1109/access.2026.3717009 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Sibei Chen und 3 weiteren Autor:innen. Die Arbeit nennt 43 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Aiming at the problems of large aspect ratio variation of slender targets, complex background interference and insufficient long-range spatial dependency modeling in remote sensing target detection, this paper proposes a remote sensing oriented object detection network named Fre-MambaNet based on frequency domain enhanced attention and improved Mamba. Firstly, we design a Spectrum-Guided Feature Enhancement (SGFE) module, which combines channel-spatial attention and lightweight frequency-domain MLP to suppress low-frequency background interference and enhance target structural features. Secondly, an Improved Standard Mamba (ISMamba) module is proposed. By replacing causal convolution with standard convolution and adopting dual-branch fusion, the bidirectional global context modeling capability for 2D remote sensing features is enhanced with low computational cost. We conduct extensive experiments on two public remote sensing datasets DOTA-v1.0 and HRSC2016. On DOTA-v1.0, our method achieves 80.21% mAP, outperforming a series of state-of-the-art methods. On HRSC2016, it obtains 90.73% mAP (VOC 07) and 98.76% mAP (VOC 12), with lower FLOPs than comparison methods. Experimental results demonstrate that Fre-MambaNet has superior performance in slender target detection and complex scene modeling.

Förderung

  • National Natural Science Foundation of China

Zitieren

APA

Chen, S., Ye, Z., Yang, Y., & He, Z. (2026). Remote Sensing Target Detection Network Based on Frequency Domain Enhanced Attention Mechanism With Mamba. IEEE Access, 14, 117453-117469. https://doi.org/10.1109/access.2026.3717009

Vancouver

Chen S, Ye Z, Yang Y, He Z. Remote Sensing Target Detection Network Based on Frequency Domain Enhanced Attention Mechanism With Mamba. IEEE Access. 2026;14:117453-117469. doi:10.1109/access.2026.3717009

BibTeX
@article{Chen2026,
  author = {Chen, Sibei and Ye, Zhiwei and Yang, Yiming and He, Zijian},
  title = {Remote Sensing Target Detection Network Based on Frequency Domain Enhanced Attention Mechanism With Mamba},
  journal = {IEEE Access},
  year = {2026},
  volume = {14},
  pages = {117453--117469},
  doi = {10.1109/access.2026.3717009}
}
Kostenlos starten →