Penerapan CNN Untuk Mendeteksi Dan Mengklasifikasikan Penyakit Daun Jagung Sebagai Pendukung Sistem Pertanian Digital

Corn is one of the most widely cultivated grains worldwide. Corn has many benefits for the human body because of its nutritional content, namely among the benefits it can reduce hypertension so that people with heart disease can take early prevention. Corn plants are very susceptible to certain leaf diseases such as northern leaf blight, common rust, and cercospora. In this study, classification was carried out to detect disease in corn leaf images using CNN. The data contained in this study will be preprocessed first, data preprocessing is very important to do to avoid unbalanced data. This study uses data augmentation to produce a better model. Data augmentation is a process in image data processing. Data augmentation is a strategy of increasing the diversity of data available for a training model, without actually collecting new data. The data used in this study was obtained from the dataset: Plant Leaf Diseases, with a total of 3852 which is divided into 4 classes, namely: healthy, cercospora, common rust and northern leaf blight. Feature extraction and classification using a convolutional neural network (CNN) using 11 models, namely: AlexNet, VGG16, VGG19, ResNet50, RestNet101, Inception-V3, DenseNet201, Xception, MobileNetV2, Inception ResNet V2, ResNet152 V2. Based on the results of the study, the best classification was obtained by DenseNet201 with an accuracy rate of 97%. ResNet152 with 96% accuracy and VGG16 get satisfactory accuracy results of 95%. DenseNet201 with augmentation data gets an accuracy rate of 97%. DenseNet201 in this study is the most suitable model for feature extraction and image classification in corn leaf disease. By looking at the accuracy results that have been carried out, it is proven that Data Augmentation applied to training data can improve accuracy results for image classification needs