Samenvatting

Aviation is indispensable for global transportation and economic growth. One promising solution to its environmental concerns is developing biofuels derived from oil seeds like Brasica carinate (B. carinata). Adopting responsible nitrogen application practices during the cultivation of B. carinata mitigate environmental degradation while optimizing yields. This study aimed to investigate plant physiological changes, represented by SPAD and Fv/Fm values, under different nitrogen treatments using machine learning models with RGB imaging and leaf spectral reflectance. 17 RGB indices, 19 morphological features were analysed. The performance of seven machine learning algorithms were compared. Using RGB indices, Single Decision Tree regression model was identified as the optimal model, achieving R2 values of 0.765 and 0.778, and RMSE of 4.149 and 4.271 for SPAD and Fv/Fm values, respectively. For morphological features, the XGBoost regression models was most optimal, with R2 values of 0.823 and 0.863, and RMSE of 3.588 and 0.048 for SPAD and Fv/Fm values, respectively. Partial Least Squares Regression models demonstrated high accuracy, achieved R2 values of 0.883 for SPAD and 0.873 for Fv/Fm, with RMSE of 0.003 and 0.008. These findings underscore the potential of machine learning in accurately predicting SPAD and Fv/Fm in B. carinata using RGB imaging and leaf spectral reflectance data.