HYPERSPECTRAL ESTIMATION OF FOXTAIL MILLET (SETARIA ITALICA) GRAIN PROTEIN CONTENTS BY USING PHOTOSYNTHETIC RATE PARAMETERS UNDER DIFFERENT PHOTOPERIODS.

To predict the grain protein content of foxtail millet in a timely, effective, and rapid manner, the study used the net photosynthetic rate of leaves as an intermediate parameter. By using hyperspectral reflectance-net photosynthetic rate of leaves-grain protein content and using the 2020 potted plant experiment data, hyperspectral data and its first derivative (1ST) data with different photoperiod treatments, the models were constructed using machine learning algorithms (SVM, PLS and BP) based on net photosynthetic rate. The results revealed that the accuracy of the SVM model was better than that of the PLS model and BP model, and the SVM model based on the first derivative (1ST) spectral reflectivity was better than the original spectral (R) reflectivity, the R2, RMSE, RPD of the modeling set and validation set were 0.988, 0.732, 0.823, 4.061, 8.697, 1.810, respectively The correlation coefficient between net photosynthetic rate and grain protein content was 0.872 and R2 was 0.719 at 10 days after anthesis. The SVM model can be used to accurately monitor the grain protein content of foxtail millet, and provide technical support for hyperspectral techniques in high-yield cultivation and Intensive farming of foxtail millet. [ABSTRACT FROM AUTHOR]