Experimental data manipulations to assess performance of hyperspectral classification models of crop seeds and other objects

Table 1 Germination data and numbers of tomato seeds included in this study

Tomato seeds from five subsamples of each of two varieties (1 and 2) were included in this study (10 samples, see also Fig. 1a). Two subsamples for each variety (variety 1: 1a and e, variety 2: 2f and j) were used as training data, and these are highlighted in bold. For all 10 tomato seed subsamples, we obtained germination results (%) from the seed company, and we performed four replicated germination tests of subsamples used as training data. Hyperspectral images of individual tomato seeds in training and validation data sets were acquired on different days, and two sets of validation data were acquired on separate days

ISSN: 1746-4811