Development of non-contact grape quality assessment method using a VIS-NIR spectroscopic device

Increasing consumer preference for crisp and sweet grapes underscores firmness, strain rate, and soluble solids content (SSC) as key quality attributes. Conventional measurement methods are labor-intensive and contribute to postharvest losses, highlighting the need for non-destructive and accessible alternatives. In this study, partial least squares regression (PLSR) models were developed to evaluate a low-cost visible-near-infrared (VIS-NIR) spectroscopy system for non-destructive grape quality assessment. SSC prediction showed consistent performance ( r = 0.82 for reflectance and r = 0.80 for transmittance), indicating reliable predictive capability. Firmness prediction achieved moderate feasibility in transmittance mode ( r = 0.60), while strain rate prediction showed comparable performance in reflectance mode ( r = 0.59). Key spectral regions (400-550, 680-700, and 920-970 nm) were identified as informative for all three attributes. These findings suggest that transmittance-based VIS-NIR spectroscopy performed better than reflectance spectroscopy for firmness prediction under the conditions investigated, whereas reflectance provided moderate predictive performance for strain rate and strong predictive performance for SSC. The proposed system, which integrates low-cost optical components with chemometric analysis, provides a cost-effective approach for non-destructive estimation of multiple grape quality attributes. Future work will focus on improving model robustness through expanded datasets and advanced data processing approaches.