A Novel Deep Learning Design of Plant Disease Recognition and Detection Using VGG19, and DenseNet121

Plant diseases pose a significant threat to global agriculture, leading to substantial yield losses and economic impact. Addressing this issue, deep learning models have emerged as powerful tools for automated plant disease detection, aiding in early and accurate diagnosis. In this study, we investigate the efficacy of two popular convolutional neural network architectures, VGG19 and DenseNet121, for plant disease detection. We propose a comparative analysis of these models, evaluating their performance in terms of accuracy, sensitivity, specificity, and computational efficiency.The examination reveals significant performance differences between plant disease detection models, particularly VGG19 and DenseNet121. VGG19 achieves a commendable validation accuracy of 88.06%, but at the cost of a significantly higher validation loss of 31.44 and a training accuracy of 85.06% with a loss of 41.20. In contrast, DenseNet121 showcases superior performance, maintaining an exceptional training loss of 6.21 while achieving a high training accuracy of 97.57%, surpassing VGG19. DenseNet121 also excels in validation, boasting a validation accuracy of 97% and a validation loss of just 2.24. These results underscore DenseNet121’s accuracy and effectiveness, positioning it as an attractive choice for tasks related to plant disease identification. Its potential to substantially contribute to plant health assessment and agricultural disease management is underscored by its remarkable accuracy in identifying disease patterns.