A Biologically Inspired Multi-Scale CNN for Edge-Based Grape Leaf Disease Detection
Engineering, Technology & Applied Science Research · 04.04.2026 · 16 (2) · 34090-34096
DOI: 10.48084/etasr.16230 ↗Kurz gefasst
Fachartikel in der Zeitschrift Engineering, Technology & Applied Science Research, online erschienen am 04.04.2026. Verfasst von C. Sushma und 1 weiteren Autor:innen. Die Arbeit nennt 31 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Grape leaf diseases can severely compromise crop yield and fruit quality, especially in regions that lack timely diagnostic support. Traditional image-based models often struggle with environmental variability and require high computational power, making them less viable for real-time field deployment. This work introduces a Biologically-inspired Multi-scale Convolutional Neural Network (BMCNN) designed to emulate hierarchical visual mechanisms for precise grape leaf disease identification. Engineered for real-time deployment on edge platforms, the architecture leverages multi-scale receptive field integration and depth-efficient convolutional modules to extract detailed lesion patterns across varied background conditions. The model was trained and validated on an augmented grape leaf dataset that encompasses multiple disease classes. Performance was compared against conventional CNNs and lightweight architectures such as MobileNet and EfficientNet. Deployment feasibility was evaluated on NVIDIA Jetson Nano and Raspberry Pi platforms. The BMCNN model achieved a classification accuracy of 91.96%, with precision and recall metrics exceeding 92% across all disease categories, delivering faster inference, reduced parameter complexity, and lower energy consumption compared to standard benchmarks. Field trials confirmed its reliability, maintaining low latency and efficient power usage. These improvements stem from a biologically inspired architecture combined with edge-optimized design. The BMCNN framework offers a scalable and interpretable solution for precision agriculture, with modular integration into IoT-based plant health systems. Future work will focus on cross-crop generalization, multimodal sensing, and adaptive learning under dynamic field conditions.
Zitieren
APA
Sushma, C., & Veerappa, B. N. (2026). A Biologically Inspired Multi-Scale CNN for Edge-Based Grape Leaf Disease Detection. Engineering, Technology & Applied Science Research, 16(2), 34090-34096. https://doi.org/10.48084/etasr.16230
Vancouver
Sushma C, Veerappa BN. A Biologically Inspired Multi-Scale CNN for Edge-Based Grape Leaf Disease Detection. Engineering, Technology & Applied Science Research. 2026;16(2):34090-34096. doi:10.48084/etasr.16230
BibTeX
@article{Sushma2026,
author = {Sushma, C. and Veerappa, B. N.},
title = {A Biologically Inspired Multi-Scale CNN for Edge-Based Grape Leaf Disease Detection},
journal = {Engineering, Technology & Applied Science Research},
year = {2026},
volume = {16},
number = {2},
pages = {34090--34096},
doi = {10.48084/etasr.16230}
}
EUVIMED