Artificial neural network based on multilayer perceptron algorithm as a tool for tomato stress identification in soilless cultivation

In this study an artificial neural network model based on multilayer perceptron (MLP) learning algorithm was developed and tested in order to estimate different type of stress in tomato crop under greenhouse conditions. Early estimation of different types of crop stress in real-time can further reduce the inputs and the energy consumption. The aim is to perform a qualitative classification of the data, depending on the type of stress (such as no stress, water stress and cold stress). To build the ML models, nine qualitative characteristics used to create the database under the different type of the crop stress. The best combination of hyperparameters which improve the current classifier to classify the different types of stress was, hidden layer sizes (70, 70, 70) and maximum number of iterations 200. The learning procedure and classification steps were written in Python language. The model was based on the 10,763 samples that were divided into two parts, one for training-validation 80% (8,610) and a second one for testing 20% (2,152). To evaluate the performance of the MLP algorithm presented in this study the Positive Predictive Values (PPV or Precision), Accuracy, Sensitivity and F1 (F1-score) were used. MLP model gave results on the validation set with 96% Accuracy, 96% Precision, 95% Sensitivity and 96% F1. Particularly, the model correctly identified 371 out of 372 samples of the cold stress plants, 1281 out of 1321 samples of the no stress plants and 403 out of 452 samples of the water stress plants.