HYBRID DEEP LEARNING WITH PCA-OPTIMIZED FEATURE SELECTION IN EARLY-STAGE PAPAYA LEAF DISEASE DIAGNOSIS

Abstract

Early detection of leaf diseases is essential to maintain crop health and improve agricultural yield. This study proposes an advanced system that uses artificial intelligence (AI) and principal component analysis (PCA) for efficient feature selection in papaya leaf disease classification. The system uses a combination of deep learning models, including VGGNet, ResNet and GoogLeNet, to extract critical features from a comprehensive dataset of healthy and diseased papaya leaf images. PCA is applied to reduce the dimensionality of the extracted features and select the most relevant features for accurate classification. The selected features are classified using linear discriminant analysis, resulting in an impressive accuracy of 96.57%. This high accuracy demonstrates the effectiveness of the proposed method in diagnosing papaya leaf diseases. In addition, the open-source nature of the system encourages reproducibility and further research, providing a valuable tool for the agricultural community. By providing a reliable and efficient solution for early disease detection, this approach will assist farmers in taking timely action, ultimately contributing to the sustainability and productivity of papaya cultivation. The integration of AI and PCA in this system marks a significant advancement in the field, highlighting its potential for wider application in agricultural disease management.