Mass and Volume Modeling of Strawberry Fruits (Arbutus unedo L) Based on Engineering Properties and Image Processing Approach

Strawberry fruits are often graded based on size, shape, and colour, but it could be more efficient to grade them using mass or volume. Therefore, the relationships between the physical characteristics of fruit and its mass or volume are needed. Present study aims to predict the mass and volume of strawberry fruit based on measured values of selected physical parameters, and develop single and multivariate regression models like linear, quadratic, power, and exponential for prediction. Further, the mass and volume models were presented under different categories such as based on dimensions (including length, width, perimeter, and projected area), volumes, and density values. Among these models, quadratic was found appropriate. The experimental data analysis showed that the R² of the quadratic model for the mass and the length, width, perimeter, and projected area were 0.893, 0.864, 0.908, and 0.909, respectively. In contrast, for the volume, R² values were 0.887, 0.872, 0.905, and 0.910 for length, width, perimeter, and projected area, respectively. The results revealed that mass and volume modeling based on the projected area were the best-fit models with better prediction accuracy. The highest coefficient of determination (R²) was obtained for mass modeling of strawberries based on measured volume as R² = 0.975. Mass modeling was recommended as the most accurate, reliable, and appropriate modeling compared to volume modeling. The findings of these study can be used as hyperparameters in automated grading system based on volume or weight of the individual fruit.