Using Optimization Algorithm to Improve the Accuracy of the CNN Model on the Rice Leaf Disease Dataset

Rice plays an essential role in daily meals. Therefore, planting and tending to play a significant role, however, the disease is an issue that needs attention and monitoring. In this work, we propose an approach to improve the accuracy of the prediction model using CNN algorithm on rice leaf dataset with 7532 samples with 5 different diseases such as bacterial blight, blast, red strip, tungro, and brown spot. This dataset uses data augmentation methods with rotations, width range shift 0.2, height shift 0.2, vertical flip, and horizontal flip. Finally, with the application of optimization models such as Adaptive Gradient Algorithm (Adagrad), Root Mean Square Propagation (RMSProp), and Adaptive Moment Estimation (Adam), the Adam optimal algorithm results in stability and accuracy. 98.06%, higher than the other 2 algorithms 72.70 and 96.86%.