The manual detection of diseases in ornamental plants is often slow, inaccurate, and prone to errors. This situation negatively impacts the quality and market value of ornamental plants in Indonesia, which possess high aesthetic and economic value. This study aims to develop a Convolutional Neural Network (CNN) model based on leaf images to detect and classify ornamental plant diseases quickly and accurately. The methodology involves collecting a dataset of leaf images categorized into five classes: bacterial, fungal, viral, pest-induced, and healthy. The CNN model was trained using the TensorFlow framework and integrated into a web-based application utilizing Laravel and FastAPI to enhance user accessibility. The results indicate that the developed CNN model achieved an accuracy of 89.67%. The implemented application is capable of early detection of ornamental plant diseases with high speed and accuracy, equipped with detection history and treatment recommendation features. This solution demonstrates a significant contribution to the application of artificial intelligence technology in agriculture, particularly in supporting the sustainable health of ornamental plants in Indonesia.