EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Explainable Hybrid Deep Learning for Tomato Leaf Disease Classification: Comparative Performance and Agricultural Decision-Support Implications

Suma G.R, Dr. Ghouse Ahmed Z

Computer and Decision Making: An International Journal · 01.11.2026 · 3 (6S) · 2816-2828

DOI: 10.1922/comdem.v3i6s.19027 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Computer and Decision Making: An International Journal, online erschienen am 01.11.2026. Verfasst von Suma G.R und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Timely identification of tomato leaf disease can be helpful for crop monitoring and agricultural assessment; however, the similar symptoms of different diseases hinder the automated recognition and interpretation. This study proposes an explainable hybrid deep learning model which fuses the MobileNetV3Large and EfficientNetB4 at the feature level, followed by multiclass classification and Gradient-weighted Class Activation Mapping (Grad-CAM). The model was tested on 30,609 images of ten tomato leaf classes and five stand-alone CNN models. The proposed hybrid obtained a training accuracy of 97.31%, a validation accuracy of 96.57%, and a test accuracy of 96.61% along with macro-averaged precision, recall and F1-score of 96.67%, 96.63% and 96.62% respectively. It achieved a test accuracy of more than 0.78 percentage points higher than the best standalone architecture, MobileNetV3Large (95.83%). In addition to the overall accuracy, the assessment also reveals the differences in recognition of the disease, including those between visually similar classes. Qualitative indications of image regions associated with predictions are provided by grad-CAM maps. Based on these findings, the classifier can be considered as an explainable information component for human supervised agricultural assessment. Experimental evaluation of decision policies, field-level management outcomes and improvements in agricultural decision quality were not conducted.

Zitieren

APA

G.R, S., & Z, D. G. A. (2026). Explainable Hybrid Deep Learning for Tomato Leaf Disease Classification: Comparative Performance and Agricultural Decision-Support Implications. Computer and Decision Making: An International Journal, 3(6S), 2816-2828. https://doi.org/10.1922/comdem.v3i6s.19027

Vancouver

G.R S, Z DGA. Explainable Hybrid Deep Learning for Tomato Leaf Disease Classification: Comparative Performance and Agricultural Decision-Support Implications. Computer and Decision Making: An International Journal. 2026;3(6S):2816-2828. doi:10.1922/comdem.v3i6s.19027

BibTeX
@article{GR2026,
  author = {G.R, Suma and Z, Dr. Ghouse Ahmed},
  title = {Explainable Hybrid Deep Learning for Tomato Leaf Disease Classification: Comparative Performance and Agricultural Decision-Support Implications},
  journal = {Computer and Decision Making: An International Journal},
  year = {2026},
  volume = {3},
  number = {6S},
  pages = {2816--2828},
  doi = {10.1922/comdem.v3i6s.19027}
}
Kostenlos starten →