Analisis Perbandingan Performa Model Klasifikasi Kesehatan Daun Tomat menggunakan arsitektur VGG, MobileNet, dan Inception V3

Abstract

One of the problems in the field of tomato farming is the spread of disease in tomato plants when there are tomato plants that are affected by the disease and are detected too late and are not treated immediately. Many studies regarding the introduction of classification in tomato plant diseases using the convolutional neural network (CNN) method. However, researchers continue to conduct deep learning on various image-based object classification tasks. In this paper, several models will be tested to classify tomato plant leaf health in order to identify diseased tomato plants. The proposed method uses the CNN approach with VGG, MobileNet, and Inception V3 architectures. The image data used comes from the plant disease classification merged (public dataset) which has many image categories used in experimental work. The experimental results show that each model has achieved accuracy performance of 98%, 93% and 88% for InceptionV3, VGG, and Mobile Net. The result is that the model with the best order in processing data is obtained by Inception V3, then VGG and Mobile Net. even so, mobileNet still has effectiveness and efficiency when running models that are far better than Inception V3 and VGG.