Strawberry Disease Detection Based on YOLOv8 and K-Fold Cross-Validation

Abstract

Strawberry plant diseases can be detected by the condition of the strawberry leaves, flowers, and fruit, but farmers still need knowledge to identify the type of strawberry disease. This study aims to develop a detection model using YOLOv8. The detection model was trained using a dataset containing 3,243 images of strawberry plant leaves, fruit, and flowers, divided into seven disease classes and one healthy plant class. This study aims to develop a more effective strawberry plant disease detection technology. The proposed method is based on YOLOv8 by applying K-Fold Cross Validation to the detection model training and applied data albumentations to produce a robust model. Based on the experimental results, it shows that the YOLOv8s model obtained the highest precision, recall, F1-score, and mean average precision values of 1.00, 0.94, 0.84, and 0.885 respectively.