Automatic method for classification of groundnut diseases using deep convolutional neural network

Groundnut is one of the most important and popular oilseed foods in the agricultural field, and its botanical name is Arachis hypogaea L. Approximately, the pod of mature groundnut contains 1–5 seeds with 57% of oil and 25% of protein content. The oil obtained from the groundnut is widely used for cooking and losing body weight, and its fats are widely used for making soaps. The groundnut cultivation is affected by different kinds of diseases such as fungi, viruses, and bacteria. Hence, these diseases affect the leaf, root and stem of the groundnut plant and it leads to heavy loss in yield. Moreover, the enlarger number of diseases affects the leaf and root-like Alternaria, Pestalotiopsis, Bud necrosis, tikka, Phyllosticta, Rust, Pepper spot, Choanephora, early and late leaf spot. To overcome these issues, we introduce an efficient method of deep convolutional neural network (DCNN) because it automatically detects the important features without any human supervision. The DCNN procedure can deeply detect plant disease by using a deep learning process. Moreover, the DCNN training and testing process demonstrate an accurate groundnut disease determination and classification result. The number of groundnut leaf disease images is chosen from the plant village dataset, and it is used for the training and testing process. The stochastic gradient decent momentum method is used for dataset training, and it has shown the better performance of proposed DCNN. From the comparison analysis, the 6th combined layer of proposed DCNN delivers a 95.28% accuracy value. Ultimately, the groundnut disease classification with its overall performance of proposed DCNN provides 99.88% accuracy.