Fecha de publicación: 17/03/2020 Fuente: Wipo "precision agriculture" The invention discloses a high-precision agricultural product classification method and system. The method comprises an agricultural product classification system establishing step, a training sampledetermining step, a modeling step, a testing step and a classification step. In the modeling step, a deep recurrent neural network is used for modeling, agricultural product categories are representedby digital numbers, names of processed agricultural products in a training set sample and classification numbers of corresponding marks are imported into the deep recurrent neural network, and modeltraining is carried out according to different types of agricultural products in an agricultural product classification system. In the classification step, whether the trained model is used as a finalclassification model or not is determined according to a result of the test step. According to the method, through deep recurrent neural network modeling, the model is applied to agricultural producttext classification, the classification precision requirement is met in a self-built agricultural product category system, the classification efficiency is high, data can be rapidly analyzed under mass data, and the method has a great effect on subsequent data processing, data application and the like.