Convolutional Neural Network Model Architecture for Rice Leaf Digital Image Identification

Rice is one of the most important staple crops in the world. Nutrition is one of the crucial things in the growth and development of rice plants. This study aims to classify the nutritional needs of rice plant leaves and determine the accuracy of the convolutional neural network (CNN) algorithm in determining nutrition in rice plants. The architectural models used in this study are VGG16, MobileNet, and Xception, using Jupyter and Google Colaboratory as tools. In this study, a dataset of 1190 images of rice leaves was used. The rice leaf image is divided into two classes, Adequate and less tested, with a ratio of 80%, training data and 10% test data, and 10% as validation data. The best accuracy results were obtained by VGG16 at 78.15% and 76.47%, MobileNet at 82.69% and 86.55%, and Xception at 82.33% and 88.24%. Meanwhile, overall, the best accuracy results were obtained from the Xception model of 88.24% with an input batch size of 32, and the tools used were Jupyter Notebook.