A LIGHTWEIGHT DEEP LEARNING MODEL FOR ACCURATE PLANT DISEASE DETECTION IN REAL-WORLD ENVIRONMENTS

Abstract

Plant diseases significantly reduce agricultural productivity and threaten global food security, particularly in regions dependent on smallholder farming. Recent advances in deep learning have shown promising results in automated plant disease detection; however, many existing models are computationally intensive and perform poorly under real-world field conditions. This research work proposes a lightweight deep learning model for accurate plant disease detection that balances high classification performance with low computational complexity. The model leverages efficient convolutional architectures and attention mechanisms to enhance feature discrimination while maintaining suitability for deployment on resource-constrained devices such as Smartphone’s edge systems. Extensive experiments demonstrate that the proposed approach achieves competitive accuracy, precision, recall, and F1-score across multiple disease classes while significantly reducing inference latency. The results confirm the model’s robustness to variations in illumination, background clutter, and symptom severity, making it practical for real-world agricultural environments. This research work contributes toward scalable, accessible, and cost-effective plant disease diagnosis systems for precision agriculture.rely heavily on routine screenings and heightened awareness of potential warning signs.

KEYWORDS

Deep Learning, Plant Disease Detection, Lightweight CNN, Real-Time Agriculture, Attention Mechanism