ANALISIS PERBANDINGAN HASIL KLASIFIKASI JENIS PENYAKIT TANAMAN TOMAT MENGGUNAKAN ARSITEKTUR MOBILENET, DENSENET121, DAN XCEPTION

Machine learning can be applied in various needs, such as image classification. Plant disease classification is essential and significantly supports the agricultural sector in this modern era. With an application capable of classifying diseases in crops, farmers can accurately identify the diseases affecting their harvest and address them more efficiently and effectively compared to traditional methods, which can be more time-consuming. This research aims to determine the best TensorFlow architecture among the three architectures used in this study, namely MobileNet, DenseNet121, and Xception, to classify 9 types of tomato plant diseases and 1 healthy tomato plant. The study concludes that DenseNet121 is the best architecture for classifying the 9 types of tomato plant diseases and 1 healthy tomato plant. During testing, the DenseNet121 model achieved an accuracy, precision, recall, and F-1 score of approximately 0.987 or 98.7%. Xception ranked second with all four metrics scoring around 0.986 or 98.6%, while MobileNet ranked last with metrics scoring approximately 0.973 or 97.3%.