<![CDATA[Early detection of plant diseases is critical for enhancing productivity and sustainability in precision agriculture, given that conventional visual inspection methods are often subjective, time-consuming, and less effective for large-scale farming. This study aims to design and develop a digital image-based plant disease detection application using deep learning to support intelligent decision-making in precision agriculture. Applying a transfer learning approach to pretrained ResNet-50 and MobileNetV2 architectures, this research utilizes a dataset of 3,000–5,000 labeled leaf images divided into training (), validation (), and testing () sets, supplemented with image preprocessing and augmentation techniques. Performance evaluation was conducted using accuracy, precision, recall, F1-score, ROC curves, and AUC metrics. Experimental results demonstrate that ResNet-50 achieved the highest performance with an accuracy of and an AUC of , while MobileNetV2 reached an accuracy of and an AUC of , both Convolutional Neural Network (CNN)-based models significantly outperforming the conventional Support Vector Machine (SVM) method, which only recorded an accuracy of . These findings conclude that CNN-based transfer learning effectively improves plant disease classification performance and provides a robust foundation for intelligent decision-support systems in precision agriculture.]]>
Copyrights © 2024
