Peanut yield estimation models using Machine Learning techniques and Google Earth Engine

Abstract

The estimation of the yield for the peanut crop can be performed by visual methods, traditional destructive methods, which are time consuming, laborious and with imprecise predictions. Thus, the objective was to use SR techniques and Artificial Neural Networks (ANN) in the development of an innovative method for yield prediction in the peanut crop. The experiments were conducted in the state of São Paulo in the 2021/2022 crop season, in 6 commercial farms in sandy and clayey areas. The RBF and MLP networks were able to estimate peanut yield with an accuracy below 1000 kg/ha. GNDVI was a better vegetation index with estimation accuracy of 238.7 and 296.54 for the RBF and MLP networks, respectively.