Mango Ripeness Classification Based on Skin Image Shape Features Using Decision Tree

Mango ripeness classification plays a vital role in the post- harvest process to ensure optimal quality during storage and distribution. This study proposes a lightweight and interpretable classification method based on Hu Moment shape descriptors extracted from mango skin images. These features were evaluated using three variants of the Decision Tree (DT) algorithm: Fine, Medium, and Coarse. The dataset was split into 90% training and 10% testing, and performance was assessed using standard metrics including accuracy, precision, recall, specificity, F-score, and training time. The results show that the Fine DT model achieved the best performance with a testing accuracy of 86.941% and the shortest training time of 4.0016 seconds, outperforming the Medium and Coarse DT variants in terms of efficiency. The findings demonstrate that Hu Moment features, despite their simplicity, are effective in representing mango shape for ripeness classification. This approach offers a promising solution for non-destructive fruit evaluation, particularly in embedded or real-time applications.