Near infrared spectroscopy for sustainable non-destructive classification and quantitative prediction of quality traits of seed size, pest damage, and quality traits in lychee (Litchi chinensis)

Lychees are highly valued in export markets if they have an aborted seed and are free from fruit borer damage. However, the traditional grading method relies on manual assessment by laborers, which is experience-dependent, time-consuming, and potentially destructive to the fruit. The objective of this study was to evaluate NIR spectroscopy for classifying premium lychee fruits based on seed size and the absence of fruit borer damage, as well as for quantifying the soluble solids content (SSC) and titratable acidity (TA). NIR spectra in the 12000-4000 cm⁻¹ range were acquired in reflectance mode at three different positions: stem end, cheek side, and tip. The classification models effectively distinguished seed size at tip side, achieving 90.58% accuracy and fruit borer damage at stem-end, resulting in 85.71% and 90.91% correct classification for normal and aborted seeds, respectively. Additionally, SSC and TA content were quantitatively assessed from the NIR spectra with R 2 values > 0.90. These results could pave the way for efficient quality control of lychees. This research advances innovative sensing technologies for agriculture while enabling sustainable production and resource-efficient food systems.