Using Optimization Algorithms to Improve The Classification Of Diseases On Rice Leaf Of EfficientNet-B4 Model

Rice is an important food source, so increasing rice yields is essential to disease management. Much related research has performed classification and disease detection on rice leaf using machine learning models. However, this study aims to synthesize data to evaluate rice leaf diseases through collected data and contribute new data sets. This data set uses optimization algorithms (RMSprop and Adam) combined with the EfficientNet-B4 model with learning rates of 0.01 and 0.001. The research showed that the optimal algorithm combined with the EfficientNet-B4 model gave high results of 93% (F1-Score) and an accuracy of 89%. The research results show the influence of optimal parameters on the models and find the most optimal parameter results.