SISTEM DETEKSI PENYAKIT DAUN BAWANG PREI MENGGUNAKAN CNN (CONVULATIONAL NEURAL NETWORK) ARSITEKTUR VGG-19

Indonesia, as an agrarian country, relies heavily on the agricultural sector as a primary pillar of its economy. In mountainous regions, leek (Allium ampeloprasum) cultivation serves as one of the main sources of income for local communities. However, farmers in Pudak Wetan Village face challenges from plant diseases such as Fusarium wilt and purple spot, which are difficult to identify in the early stages. The disease detection process, which is still carried out manually, has proven to be less accurate, time-consuming, and prone to errors, potentially leading to reduced yields and economic losses. To address these challenges, this study was conducted to develop a system capable of automatically detecting diseases in leek plants by employing the Convolutional Neural Network (CNN) method using the VGG-19 architecture. This architecture, with its deep and stable network, is effective for image recognition tasks. CNN enables the extraction of essential features from leaf images and classifies disease types such as Fusarium wilt and purple spot, providing outputs in the form of accuracy metrics, treatment recommendations, and disease distribution graphs. The system is designed as a web-based application, allowing users to upload leaf images and receive classification results along with supporting visualizations. Image augmentation techniques were also applied to enhance the model’s generalization ability and prevent overfitting. Evaluation results demonstrated that the model achieved an F1-score of 1.00 for Fusarium wilt, 1.00 for purple spot, and 1.00 for healthy leaves. Furthermore, both macro average and weighted average scores also reached 1.00, indicating highly accurate and consistent detection performance. The outcome of this research is a system that can be used for early disease detection and assist farmers in improving both the quality and quantity of agricultural production.