Comparative Analysis of the Performance of VGG16 and ResNet50 Architectures in Multi-Class Classification of Rice Plant Diseases Based on Convolutional Neural Networks (CNN)

Abstract

Rice plant diseases significantly affect crop productivity and food security, making early and accurate disease detection essential for effective agricultural management. Recent advances in deep learning, particularly Convolutional Neural Networks (CNN), have demonstrated strong potential in image-based plant disease classification. This study presents a comparative analysis of the performance of VGG16 and ResNet50 architectures for multi-class classification of rice plant diseases using CNN-based approaches. A dataset of rice leaf images representing multiple disease classes and healthy conditions was collected and preprocessed through image resizing, normalization, and data augmentation to enhance model generalization. Both pre-trained models were fine-tuned using transfer learning to adapt them to the rice disease classification task. Model performance was evaluated using standard metrics, including accuracy, precision, recall, F1-score, and confusion matrix analysis. The experimental results show that both architectures achieve high classification performance; however, ResNet50 demonstrates superior accuracy and better generalization capability compared to VGG16, particularly in handling complex disease patterns and intra-class variations. Meanwhile, VGG16 offers a simpler architecture with faster convergence and lower computational complexity. The findings of this study provide insights into the selection of appropriate CNN architectures for rice plant disease classification and support the development of intelligent decision support systems in precision agriculture.