Prediction of garlic clove volume and mass using a depth camera and machine learning models

Accurately predicting the volume and mass of garlic cloves is essential for precision in agricultural operations, such as sorting and grading. In this study, the ellipsoid volume equation and machine learning models—Support Vector Machines (SVM), Random Forest, Gradient Boosting, and k-Nearest Neighbors (kNN)—to predict garlic clove volume and mass using length, width, height, and mass data. The SVM model excelled in volume prediction with an R² of 0.786 and a MAPE of 0.084, while the Random Forest model achieved the highest accuracy for mass prediction, with an R² of 0.849 and a MAPE of 0.098. Depth cameras further enhanced model performance by providing precise dimensional data. These findings underscore the potential of combining depth cameras with machine learning to achieve accurate, non-contact predictions of volume and mass. This approach presents promising applications for enhancing automation and quality control in agricultural systems.