Optimization methods for soybean crop disease classification: A comparative study

Abstract

India's most widely utilized food crop is soybean, and deep learning techniques are frequently used in forecasting and classification tasks. The minute scenario shows that the classification of the soybean crop diseases is a well-used machine learning technique with the help of images. But the proposed work, for the first time, combines soybean physic crop properties, weather properties, and deep learning techniques for classification. As a result, Random Forest and Support Vector Machine classification algorithms are utilized and the accuracy is compared with and without feature selection. Disease classification is compared using deep learning techniques like Recurrent Neural Networks, Convolutional Neural Networks, and Multi-Layer Perceptrons, along with optimization techniques like Adam, RmsProp, and AdaGrad. Results indicate that the farmers can predict soybean crop disease based on weather and the physical crop properties, hence taking preventive action.