EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

DELG Net: A Dual-Stream Cross-Attention Framework for Automated Nutrient Deficiency Detection in Mulberry Leaves

S. Raghavendrachar, Rekha B. Venkatapur, V. Karthik

Engineering, Technology & Applied Science Research · 04.04.2026 · 16 (2) · 33783-33789

DOI: 10.48084/etasr.17517 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Engineering, Technology & Applied Science Research, online erschienen am 04.04.2026. Verfasst von S. Raghavendrachar und 2 weiteren Autor:innen. Die Arbeit nennt 28 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Nutrient deficiencies in mulberry leaves play a decisive role in sericulture, since they directly affect the growth of silkworms and the production of silk. Traditional visual observation by professionals is subjective, laborious, and cannot be used in massive monitoring. Although generic Convolutional Neural Network (CNN) models are frequently applied to this task, they often struggle to capture the fine-grained spatial and textural details essential for identifying specific nutrient stress. To address these drawbacks, this paper presents DELG Net, a biologically-related two-stream convolutional model that conjoins global and regional feature processing. Macroscopic color and contour features are captured by the global stream, and vein and texture patterns are highlighted with high-pass filtering by the local stream. A cross-attention fusion model dynamically increases the contribution of the two streams according to the prominence of the symptoms, and feature weighting can be adjusted adaptively and contextually. Experiments on a curated dataset of mulberry plant leaves show that DELG Net achieves better precision and interpretability than state-of-the-art CNNs, with a total classification accuracy of 95.4% and consistent results in all three classes of nitrogen, potassium, and phosphorus deficiencies. The proposed model can be used to monitor nutrient stress in real-time and on a large scale to increase the efficiency of silk production.

Zitieren

APA

Raghavendrachar, S., Venkatapur, R. B., & Karthik, V. (2026). DELG Net: A Dual-Stream Cross-Attention Framework for Automated Nutrient Deficiency Detection in Mulberry Leaves. Engineering, Technology & Applied Science Research, 16(2), 33783-33789. https://doi.org/10.48084/etasr.17517

Vancouver

Raghavendrachar S, Venkatapur RB, Karthik V. DELG Net: A Dual-Stream Cross-Attention Framework for Automated Nutrient Deficiency Detection in Mulberry Leaves. Engineering, Technology & Applied Science Research. 2026;16(2):33783-33789. doi:10.48084/etasr.17517

BibTeX
@article{Raghavendrachar2026,
  author = {Raghavendrachar, S. and Venkatapur, Rekha B. and Karthik, V.},
  title = {DELG Net: A Dual-Stream Cross-Attention Framework for Automated Nutrient Deficiency Detection in Mulberry Leaves},
  journal = {Engineering, Technology & Applied Science Research},
  year = {2026},
  volume = {16},
  number = {2},
  pages = {33783--33789},
  doi = {10.48084/etasr.17517}
}
Kostenlos starten →