RNN

Land cover classification is the key challenging area in agriculture, it involves recognizing the type and quality of land. In the past, a lot of applications were based on mono-​temporal observation. Mono-​temporal methods are dependent on some factors like weather.

To solve the problems related to RNN, a model known as NARX is introduced, where NARX means nonlinear autoregressive model process with exogenous input. In this method, previous prediction values are considered as input, and present and previous values are considered as exogenous input. The system not only judges the independent inputs but also the previous response of the system which makes the system more powerful. Using NARX model another model, NARXNN, is developed for estimation of time series of leaf area index (LAI). Kurumatani (2018) proposed a technique to forecast the price of agricultural product using RNN.

RNN is also used to predict the weather. Biswas et al. (2014) designed three models for prediction of weather: nonlinear autoregressive with exogenous inputs neural network (NARX NN), case-​based reasoning model and segment case-​based reasoning model. Palangpour et al. (2016) produced a model to identify the location of animals in the forest. In this model, particle swarm optimization algorithm combined with RNN model was used, the results obtained by this model contain less errors.