A HYBRID DEEP LEARNING APPROACH FOR ACCURATE PLANT LEAF DISEASE DETECTION
DOI: https://doi.org/10.65725/JCISE/1/2/003
JOURNAL OF COMPUTATIONAL INTELLIGENCE SCIENCE AND ENGINEERING (JCISE)
ISSN: 3107-8168
Volume 1 Issue 2, October 2025

Abstract

This research addresses the critical challenge of plant disease identification, focusing on leaf-based detection through a hybrid deep learning approach. A deep convolutional neural network (CNN) based on the ResNet50 architecture is employed for automated feature extraction from plant leaf images, while a multi-layer perceptron (MLP) is utilized for the final disease classification. The CNN model is trained using the Adam optimizer, which improves training stability and generalization by decoupling weight decay from gradient updates. The methodology includes image acquisition, preprocessing, and data augmentation to enhance dataset diversity and model robustness. Implemented using MATLAB 2019b and Python, the proposed approach achieves training, testing, and validation accuracies exceeding 97%. These results demonstrate the effectiveness of integrating deep CNN architectures with MLP classifiers for precision agriculture and automated plant disease diagnosis.

Authors: Nehru Revathy, and A. Meenakshi

Keywords: Plant Disease Detection; Deep Learning; Deep CNN; ResNet50; MLP; Adam Optimizer; Precision Agriculture