Rice disease classification based on leaf damage using deep learning

Rice is the staple food of the Indonesian people. Food security is an absolute thing to do today to reduce rice imports. Various efforts were made to improve seed quality, resistance to pests, nutrition, and so on. It helps to Analysis identify diseases that may arise in rice plants. Efforts were made using the image of leaves affected by the disease and then analyzed using a deep learning algorithm. The architecture used is google net and customnet. The selection is based on a good level of accuracy and the computational time required to obtain the model. the convex hull algorithm helps to find the focus of disease objects on rice leaves. Data augmentation increases the variation in the amount of data and reduces unbalanced datasets. The results obtained are by using this algorithm, accuracy is obtained 80.94%, and the average computation time is three minutes and twenty seconds. Error calculation of classification are MSE 0.4227, RMSE 0.6501, and MAE 0.2555.