Abstract
Indian potato (Solanum tuberosum) grading is important for consumers as well as for the food industries. An automised classification system must be developed to enhance the quality of the grading and speed up the potato sorting process. As per standards and export criteria, potatoes should be divided into three groups (small, medium, and big) based on their estimated weight. This study presents the potato image-based mass modelling system. The image-based weight prediction and classification are obtained through the application of a unique dimensional analysis (DA) technique. It uses several physical attributes related to potatoes, such as size (1D, 2D, and 3D), shape, and gravimetric properties. The purpose of developing the self-built image-based system was effective automation. With regard to the training data set, the correlation coefficient (R2) between the DA predicted and actual weight of potatoes was 0.9975 for the entire set and 0.9974, 0.9990, and 0.9995 for classes A, B, and C. The DA algorithm is the most successful algorithm for predicting weight and classifying it according to grades. DA effectively works as a predictor and classifier with an efficiency of more than 99.90%. Hence, it is projected for effective automation.
