EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

DAAF: Dual-stream adaptive attention fusion with distribution alignment for remote sensing object detection

Mo Zhou iD, Yue Zhou, Kai Song iD

PLOS One · 2026

DOI: 10.1371/journal.pone.0353309 ↗

Kurz gefasst

Fachartikel in der Zeitschrift PLOS One, 2026. Verfasst von Mo Zhou und 2 weiteren Autor:innen. Beteiligte Einrichtung: School of Information Science and Engineering, Shenyang Ligong University, Shenyang, Liaoning, China. Hauptthemen laut MEDLINE: Remote Sensing Technology, Image Processing, Computer-Assisted. Die Arbeit nennt 69 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: General Project of Basic Scientific Research of Liaoning Provincial Department of Education.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Remote sensing object detection faces three challenges: extreme scale variation, arbitrary rotational orientations, and complex intermingled backgrounds. Although fusion of Convolutional Neural Networks (CNNs) and Transformers combines spatial precision with global modeling, it faces two limitations: (i) feature distribution misalignment between modalities, and (ii) ineffective exploitation of complementary discriminative strengths. To address these issues, we introduce DAAF (Dual-stream Adaptive Attention Fusion), a three-stage fusion framework that systematically aligns and integrates CNN and Transformer features. DAAF comprises three components: (1) The Unified Feature Distribution Calibration (UFDC) module applies instance normalization to align feature distributions, reducing Kullback-Leibler (KL) divergence by 94%; (2) The Heterogeneous Channel-wise Synergistic Selection (HCSS) module employs independent attention pathways for each stream, validated by zero overlap in top-10 channel weights between CNN and Transformer branches; (3) The Spatially Adaptive Discriminative-Preserving Gate (SADPG) module employs pixel-wise gating to adaptively balance fused and single-branch features. Experiments on three benchmarks (NWPU VHR-10, DOTA v1.0, DIOR) show consistent improvements averaging +2.05 percentage points (pp) in mean Average Precision (mAP) over the Add Fusion baseline, with substantial gains on structurally complex categories in NWPU (Bridge: + 17.36pp, Vehicle: + 13.18pp). These improvements are achieved with only 206K additional parameters (0.37% increase over baseline) and 7.2% latency rise compared to the baseline fusion method, demonstrating favorable efficiency for practical deployment.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Remote Sensing TechnologyImage Processing, Computer-AssistedConvolutional Neural NetworksDetection AlgorithmsAlgorithms

Autor:innen und Institutionen

  • School of Information Science and Engineering, Shenyang Ligong University, Shenyang, Liaoning, China.
  • School of Big Data, Liaoning Finance and Trade College, Huludao, Liaoning, China.

ORCID-Profile: Mo Zhou, Kai Song

Förderung

  • General Project of Basic Scientific Research of Liaoning Provincial Department of Education

Zitieren

APA

Zhou, M., Zhou, Y., & Song, K. (2026). DAAF: Dual-stream adaptive attention fusion with distribution alignment for remote sensing object detection. PLOS One. https://doi.org/10.1371/journal.pone.0353309

Vancouver

Zhou M, Zhou Y, Song K. DAAF: Dual-stream adaptive attention fusion with distribution alignment for remote sensing object detection. PLOS One. 2026. doi:10.1371/journal.pone.0353309

BibTeX
@article{Zhou2026,
  author = {Zhou, Mo and Zhou, Yue and Song, Kai},
  title = {DAAF: Dual-stream adaptive attention fusion with distribution alignment for remote sensing object detection},
  journal = {PLOS One},
  year = {2026},
  doi = {10.1371/journal.pone.0353309}
}
Kostenlos starten →