Multispectral imaging allows a quick and non-destructive determination of the seed viability of Vicia sativa

Fast, non-destructive and low-cost detection of seed viability is important for seed storage and agricultural production. In this study, multispectral imaging (MSI) was applied to identify the seed viability of three cultivars of Vicia sativa. The results showed that both linear discriminant analysis (LDA) and support vector machine (SVM) for specific cultivars showed a high accuracy in distinguishing viable and non-viable seeds. Seed imbibition favoured seed viability discrimination in three cultivars with either LDA or SVM. However, the cultivar-specific transferability is quite low among cultivars regardless of their imbibition state in our study. These results suggested that MSI could be developed as a fast and cost-effective seed viability classification method for any specific cultivar of Vicia sativa. However, the low transferability of the discrimination model among cultivars may limit this technology application in more general scenarios.