Secure and Sovereign Deep Learning for Plant Disease Detection: A Transfer Learning Approach with DenseNet121
Procedia Computer Science · 2026 · 282 · 389-399
DOI: 10.1016/j.procs.2026.05.086 ↗Kurz gefasst
Fachartikel in der Zeitschrift Procedia Computer Science, 2026. Verfasst von Pradeep Kumar Mahapatro und 4 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 21 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).
Abstract (Originalsprache)
The incorporation of digital sovereignty concepts into AI systems is crucial for ethical data usage and technological independence, especially in agriculture. This study examines how these principles support deep learning–based plant disease monitoring systems. We used several convolutional neural networks with transfer learning on the PlantVillage dataset containing images of various plant diseases, using AlexNet, VGG, ResNet, and DenseNet as pre-trained models. Our objective was to build a plant disease detection system supporting both accuracy and ethics. We evaluate data sovereignty compliance and performance indicators (accuracy, F1-score, precision, recall). We confirmed that the performance of the model and the principle of data sovereignty are highly correlated. We utilized the different models, such as AlexNet, DenseNet121, ResNet50, CNN, VGG16, as well as the basics of NN. We found the DenseNet model is suitable for soil fertility in FL settings. DenseNet121 achieved 99.657% accuracy and an F1-score of 0.99656, enabling secure disease identification while preserving data ownership. Overall, this study shows that transfer learning can include digital sovereignty criteria without affecting analytical performance. DenseNet121 achieved 99.657% accuracy and an F1-score of 0.99656, enabling secure disease identification while preserving data ownership. Overall, this study shows that transfer learning can include digital sovereignty criteria without affecting analytical performance.
Zitieren
APA
Mahapatro, P. K., Panigrahi, R., Padhy, N., Behera, S., & Arangi, D. (2026). Secure and Sovereign Deep Learning for Plant Disease Detection: A Transfer Learning Approach with DenseNet121. Procedia Computer Science, 282, 389-399. https://doi.org/10.1016/j.procs.2026.05.086
Vancouver
Mahapatro PK, Panigrahi R, Padhy N, Behera S, Arangi D. Secure and Sovereign Deep Learning for Plant Disease Detection: A Transfer Learning Approach with DenseNet121. Procedia Computer Science. 2026;282:389-399. doi:10.1016/j.procs.2026.05.086
BibTeX
@article{Mahapatro2026,
author = {Mahapatro, Pradeep Kumar and Panigrahi, Rasmita and Padhy, Neelamadhab and Behera, Sadananda and Arangi, Dasaradha},
title = {Secure and Sovereign Deep Learning for Plant Disease Detection: A Transfer Learning Approach with DenseNet121},
journal = {Procedia Computer Science},
year = {2026},
volume = {282},
pages = {389--399},
doi = {10.1016/j.procs.2026.05.086}
}
EUVIMED