For the evaluation of disease susceptibility and overwintering performance in pasture cultivar breeding, 88 RGB vegetation indices, including GRVI, were compared with the breeders’ scores from aerial images taken by an unmanned aerial vehicle (UAV) equipped with a common RGB camera. In the disease severity evaluation, 11 vegetation indices (GRVI, GmR, GmR_2, MGVRI, VARI, GI, RGRI, ExGR, ExR_2, SAVI, and SAVI_2) showed a high correlation (mean of R2 ≥ 0.5) with the breeders’ score. Most of the vegetation indices were vegetation indices that were less affected by B. Eight vegetation indices (g, ExG, ExG_2, GLI, CIVE, CIVE_2, RGVBI, and RGBVI) showed a high correlation with the breeders’ score in the overwinterability evaluation. The 8 vegetation indices that correlated well with the overwinterability evaluation were vegetation indices that used all the RGB colors. In addition, a comparison of breeders’ scores in breeding trials of oats, Italian ryegrass, Sudan grass, perennial ryegrass, and red clover with rG, a relative indicator of GRVI, showed a high correlation in the evaluation of grass vigor and degree of winter kill. On the other hand, correlations were low for grass shape and degree of lodging. In addition, regardless of the type of trait, the correlation was low in tests where the evaluation by the breeder itself was difficult. These results suggest that the rG evaluation is suitable for traits evaluated by the size of the plant area in the test area or by the color tone of the entire test area in aerial images taken from the sky.
View full abstract