Cybernetic Systems for Groundnut Crop Disease Detection Through Generative Adversarial Networks

Plant pests and diseases spread more quickly and, if left unchecked, can result in widespread damage to agricultural productivity and the environment. Pests and diseases also have a substantial impact on the quality and quantity of food production. Manually observing plant diseases is really difficult. It necessitates a vast knowledge of plant diseases, a large amount of labour, and a lengthy time frame to identify the diseases. To meet the food demand of the rapidly growing global population, it is necessary to introduce smart farming techniques and practices for sustainable agriculture. The agriculture sector utilizes the latest technologies like agricultural cybernetics, machine learning, and the Internet of Things for sustainable agriculture to enhance food productivity and food security for future generations. In this research, a novel Taylor Cat Swarm Optimization (TCSO) algorithm was developed to classify different groundnut diseases with higher classification accuracy. The SegNet architecture is used for image segmentation and the key features are extracted from the segmented image for diagnosing crop disease. The proposed Taylor-CSO algorithm is used to train the Conditional Generative Adversarial Networks (CGAN) to detect crop diseases accurately with a limited dataset. Also, it provides faster convergence when compared with GAN and generates a controlled output image for the given conditional labels. The proposed algorithm achieves an accuracy of 92.6% in detecting groundnut crop diseases when compared with the MFCN of 64%, the deep learning-NN of 76%, and the CNN of 86%. Finally, the performance results prove that the proposed algorithm outperforms other algorithms in terms of accuracy, sensitivity, and specificity.