| dc.description.abstract | Monstera ornamental plants (Monstera spp.) are popular among plant lovers due to their unique and aesthetically pleasing leaf shapes. However, these plants are susceptible to various diseases that can inhibit growth, reduce visual appeal, and even lead to plant death. Manual disease identification is a challenge for cultivators due to limited knowledge of plant diseases characteristics and the similarity of symptoms among different diseases, making it difficult for cultivators to determine the right treatment steps. Therefore, an method is needed that is able to classify diseases on monstera leaves quickly and accurately by utilizing color and texture images on monstera leaves. This research applies the EfficientNet architecture to classify diseases on monstera leaves. The dataset used amounted to 400 leaf images consisting of 280 training data, 80 validation data and 40 testing data. The entire dataset were categorized into four classes: healthy leaves, fungal leaf spots, bacterial leaf spots and anthracnose. The leaf image dataset was collected through direct capture using a smartphone camera, and then goes through several pre-processing stages such as cropping, resizing, and augmentation. The results shows that the EfficientNet model is able to achieve 90% accuracy in classifying on monstera leaves disease. These findings indicate that the developed system can accurately identifying the type of monstera leaf disease well. Afterward, the model was implemented into an android-based application. | en_US |
