A Federated Learning Strategy for Secure and Localized Detection of Diseases in Plant Using Leaf Imagery
Procedia Computer Science · 2026 · 283 · 123-131
DOI: 10.1016/j.procs.2026.06.101 ↗Kurz gefasst
Fachartikel in der Zeitschrift Procedia Computer Science, 2026. Verfasst von Ishana Attri und 2 weiteren Autor:innen. Die Arbeit nennt 16 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).
Abstract (Originalsprache)
Timely identification of plant diseases plays an essential role in sustaining agricultural productivity and ensuring food secu- rity. Traditional deep learning methods for recognizing these diseases often rely on pooling data in a centralized manner, which raises issues around privacy, data ownership, and connectivity—especially for farmers in remote or rural areas. To tackle these challenges, this study proposes a federated learning framework designed to protect data privacy while employing a lightweight convolutional neural network tailored for edge-device deployment. We trained our model using two publicly available datasets: the Apple Leaf Disease dataset, which contains 9,714 images spanning four disease categories, and the Mango Leaf Disease dataset, comprising 1,405 images across three categories. To simulate real-world agricultural diversity, experiments were carried out with non-identically distributed data across ten clients. Our federated CNN achieved strong performance, with 86.3% accuracy, 86.0% precision, 86.5% recall, and an F1-score of 86.2%. These results outperform baseline centralized and federated models while main- taining strict data privacy. The framework offers promising potential for on-site, real-time disease diagnosis, even within limited resource and connectivity environments. Future efforts will focus on enhancing privacy safeguards and deploying the system in real agricultural settings.
Zitieren
APA
Attri, I., Awasthi, L. K., & Sharma, T. (2026). A Federated Learning Strategy for Secure and Localized Detection of Diseases in Plant Using Leaf Imagery. Procedia Computer Science, 283, 123-131. https://doi.org/10.1016/j.procs.2026.06.101
Vancouver
Attri I, Awasthi LK, Sharma T. A Federated Learning Strategy for Secure and Localized Detection of Diseases in Plant Using Leaf Imagery. Procedia Computer Science. 2026;283:123-131. doi:10.1016/j.procs.2026.06.101
BibTeX
@article{Attri2026,
author = {Attri, Ishana and Awasthi, Lalit Kumar and Sharma, T.P.},
title = {A Federated Learning Strategy for Secure and Localized Detection of Diseases in Plant Using Leaf Imagery},
journal = {Procedia Computer Science},
year = {2026},
volume = {283},
pages = {123--131},
doi = {10.1016/j.procs.2026.06.101}
}
EUVIMED