Klasifikasi Hama Ulat Pada Citra Daun Sawi Berbasis Convolutional Neural Network Dengan Model Xception

Abstract

Mustard plants belong to the Cruciferae (cabbage) family. Mustard cultivation cannot be separated from disease and pest threats. The main pests and diseases in mustard plants are caper, armyworms, grasshoppers, and snails. This research aims to classify caterpillar pests in mustard leaf images using the Xception architecture model based on Convolutional Neural Network. The mustard leaf dataset used consisted of 500 images of mustard leaves without pests and 500 images of mustard leaves containing caterpillar pests. The preprocessing process includes image cropping, image normalization, image augmentation and median filtering processes. The classification stage uses the Convolution Neural Network method with the Xception architecture. The results of classification using the Xception architectural model with trials on training data with 600 training data, 200 test data and 200 validation data produced the highest accuracy value of 96%, sensitivity value of 96% and specificity value of 97%.