Klasifikasi Penyakit Pada Daun Anggur Menggunakan Convolutional Neural Network Dengan Arsitektur Efficientnet

Abstract

The grapes themselves can be consumed and can even be processed into various products so that the grapes can provide added value. This can create opportunities to increase wine production so as to increase the income of grape growers and promote the development of the wine processing industry. Factors that influence the occurrence of diseases in grapevines are environmental conditions, growing media, animals, fungi, and others. Diseases on grape leaves can interfere with the growth process of grapevines so that the grapevines are unable to produce maximum fruit. Convolutional Neural Network(CNN) has a different approach from traditional machine learning which seeks and selects among the many feature extractions to be assigned. CNN learns feature extraction through image patterns automatically from the training process. The results of the Confusion matrix used 1222 training data which were divided into 2 classes, namely Fake Flour with 585 images and Fungus Powder with 637 images. Training the EfficienNet-B0 model on grape leaf disease which produces the highest accuracy value is foundinexperiment 2 with the Adam optimizer and neurons in dense totaling 512 resulting in an accuracy value of up to 76.2%.