<![CDATA[Diseases in mustard leaves can reduce productivity if not detected early. This study aims to develop and evaluate a disease classification system for mustard leaves using Convolutional Neural Network (CNN) architectures, specifically Xception and VGG19, while comparing their performance in terms of accuracy and computational efficiency. The mustard leaf image dataset undergoes preprocessing before being used for model training and testing. Experimental results show that Xception achieves the highest validation accuracy of 99% with better loss stability compared to VGG19, which attains 94.50% accuracy but exhibits greater fluctuation. In terms of time efficiency, VGG19 reaches optimal accuracy faster and completes the training process in 42 seconds, whereas Xception requires more epochs and a training time of 50 seconds. Therefore, Xception is recommended for classification tasks that demand high accuracy and stability, while VGG19 is more suitable for rapid detection with a slight trade-off in accuracy stability.]]>
Copyrights © 2025
