Hybrid Deep-Machine Learning Based Performance Comparison for Soybean Plant Disease Identification

The seeds of soybeans, which are the most important oil crop in the world due to their high protein and oil content, are the main reasons they are farmed. However, plant diseases significantly reduce agricultural productivity and forestry and cause financial losses. In this study, we used a hybrid of convolutional neural network with a support vector machine, a random forest, and a decision tree to detect and classify diseases in soybean plants. The model was trained using images taken from kaggle.com. We developed the architecture of hybrid CNN-SVM, CNN-RF, and CNN-DT model by replacing the fully connected layer with the soft-max, which is the top layer of the CNN model, with an SVM, RF, and DT classifier. The feature maps from the soybean leaf dataset are flattened into a one-dimensional array and used as inputs for the SVM, RF, and DT. The experimental results show that the hybrid CNN-RF outperforms the hybrid CNN-SVM and CNN-DT based on test accuracy and validation accuracy. In general, the hybrid CNN-RF is suitable for the detection and classification of soybean disease.