Designing of Lightweight Deep Learning Framework for Plant Disease Detection

Plant leaf diseases pose significant challenges to agricultural productivity and food security. To address this issue, we propose a grid search tuning method to optimise convolutional neural network (CNN) parameters for plant disease identification. Our study focuses on the lightweight CNN model for efficient deployment in agricultural settings. Imaging technology and advanced Artificial Intelligence (AI) techniques offer promising solutions for early disease detection. By automating image analysis, we aim to develop a cost-effective system for identifying leaf diseases. We optimise the CNN model for improved performance by leveraging grid search hyperparameter tuning. Our proposed system achieves significant compression, making it suitable for real-world deployment. We evaluate the system using PlantVillage and cotton leaf datasets. The proposed model size for 3-layers and 4-layers is 1.81 MB and 2.77 MB. The proposed model achieves an accuracy of 99.66% and 98.46% and a loss of 0.01389 and 0.06411 for the PlantVillage and Cotton leaf disease datasets.