Classifying Rice Bacterial Panicle Blight by Combining YOLOv5 Model and Convolutional Neural Network

Rice Bacterial Panicle Blight is a highly destructive and rapidly spreading disease on rice, reducing rice yield. Therefore, this study has identified the diseased rice variety on the plant, from which it is possible to assess the extent of damage caused by the disease on rice. Because of conducting to collect the deceased rice in the field, healthy rice was mixed with diseased rice. Therefore, the study was carried out using the YOLOv5 model to detect all rice seeds in images. Then, this research performed the separation of healthy and diseased rice seeds into 2 data sets. Including 417 photos of healthy rice seeds and 314 photos of diseased rice seeds. Using the CNN model for classification, the accuracy on the train is almost 100%, and 98.68% on the test set.