Penerapan metode CNN pada Tanaman Daun Pisang dengan Model Arsitektur VGG16

The application of the Convolutional Neural Network (CNN) method in the recognition and classification of banana leaf images has shown great potential in the field of smart agriculture. This study focuses on the use of the VGG16 architectural model to analyze and classify banana leaf images to detect diseases and plant health conditions. VGG16, which consists of 16 layers, is known for its depth and ability to effectively capture image features. In this study, a dataset of banana leaf images was collected and processed to be trained using VGG16. The dataset used was taken from Kaggle under the title "Banana Leaf Spot Diseases (BananaLSD) Dataset," consisting of 2537 banana leaf images. This dataset is divided into four classes: Sigatoka (a banana leaf disease caused by the fungus Pseudocercospora fijiensis), Pestalotiopsis, Cordona (spots), and Healthy (healthy leaves). In this study, the VGG16 model was used to train and classify these images. Experimental results showed that the VGG16 model achieved an accuracy of 95.77%, precision of 96%, recall of 94%, and F1-score of 95%. The implementation of CNN with the VGG16 architecture is expected to aid in the automatic diagnosis of banana leaf diseases, providing an effective tool for farmers to monitor plant health in real-time, and opening opportunities for the application of similar technology to various other types of plants.