Hyperparameter Tuned Hybrid Convolutional Neural Network (H-CNN) for Accurate Plant Disease Classification
Shubham Sharma, Manu Vardhan
Abstract
Accurate identification and classification of plant diseases are essential for maintaining crop health and maximizing yields. Deep learning techniques have shown great promise in this field, and this study presents a Hybrid Convolutional Neural Network (H-CNN) model for plant disease classification. The proposed model architecture incorporates multiple Convolutional (Conv), and Fully Connected (FC) layers, which are trained on a publicly available dataset of plant leaf images. The model utilizes both visual and spectral features of plant leaves, contributing to its high accuracy in classifying various plant diseases. The visual features are obtained using a pre-trained VGG-16 network, while the spectral features are extracted using Principal Component Analysis (PCA). The hyperparameters for the CNN layers are optimized using a grid search algorithm. Our proposed model achieves impressive performance, outperforming state-of-the-art methods with an accuracy of 99.2% on the test dataset. The resulting H-CNN model provides a robust and accurate tool for plant disease detection and classification, with potential applications in precision agriculture and plant disease management. This study significantly contributes to the field of plant disease classification and has practical implications for agricultural researchers and practitioners.