LONG BEAN LEAF DISEASE IDENTIFICATION SYSTEM BASED ON MOBILE USING CONVOLUTIONAL NEURAL NETWORK (CNN) METHOD

Long beans (Vigna unguiculata subsp. sesquipedalis), have high nutritional value, besides long beans also have a significant role in the economy of farmers in Indonesia. However, the productivity of this plant is often hampered by various diseases that attack the leaves, which can result in a decrease in the quantity and quality of the harvest. This study has succeeded in developing a Convolutional Neural Network (CNN) model with the ResNet-50 architecture to identify six types of diseases in long bean leaves. The dataset used consists of 2,316 images, divided into training data (80%), validation (15%), and testing (5%). The ResNet-50 model, which consists of 50 layers, applies the transfer learning technique by not training the first 35 layers using a specific dataset, but utilizing weights from ImageNet. Training for 100 epochs produces high accuracy, namely 98.3% for training data, 98.4% for validation data, and 98.7% for testing data. Evaluation using Confusion Matrix, Precision, Recal and F1 Score shows very good performance without prediction errors. The final result of this research is a mobile-based software system that can diagnose diseases quickly and accurately, which can help farmers take appropriate action, and support sustainable agriculture in Indonesia.