<![CDATA[Measuring phenotypic parameters is important in evaluating the productivity of sugarcane. Existing manual measurements are considered less efficient, so a better alternative method is needed. This research aims to explore the potential of using multispectral drone imagery and machine learning to estimate phenotypic parameters of sugarcane plants that are efficient, accurate, inexpensive, and support sustainable agricultural practices. Spectrum data captured by drones, namely Green, Red, RedEdge and NIR are used as inputs to estimate phenotypic parameters including brix value, number of stands, stem diameter, and plant height. Based on the results of machine learning model development, the ANN algorithm model is most effective in predicting Brix Value with R2 0.74 and RMSE 0.06 and number of stands with R2 0.68 and RMSE 2.13. All models could not predict stem diameter and plant height well. The best model to predict plant height was obtained by RF algorithm with R2 0.53 and RMSE 14.09. SVR algorithm was the best model to predict plant diameter with R2 0.39. and RMSE 0.49. This indicates that the effectiveness of an algorithm depends on the specific parameter being predicted and there is no dominant algorithm for all phenotypic parameters. Keywords: Machine learning, Multispectral drone imagery, Phenotypic parameter, Plant productivity, Sugarcane.]]>
Copyrights © 2024
