Abstract
This paper presents a reliable image-based mass modeling system for the Indian potato (Solanum tuberosum) system, utilizing various physical properties and mathematical models to determine the impact and correlation with potato weight. The system uses an atomized image-based computer vision system to measure potato weight and develops parametric and non-parametric models for prediction. Four models were used linear, quadratic, S-curve, and power. All properties were statistically significant at the 1% probability level. The best models were based on geometric mean diameter and projected area, with R2 coefficients of 0.9702 and 0.9923, respectively. An effective grading system offers several significant benefits to both potato farmers and the potato processing industries. The outcomes of the presented work will help for higher profitability, better quality products, and maintaining consistency, operational efficiency, data driven decision making, improving sustainability, and waste reduction.
