Shuffled shepherd social optimization based deep learning for rice leaf disease classification and severity percentage prediction

• Published in: Concurrency and computation Vol. 35; no. 4
• Main Authors: Daniya, Thavasilingam, Srinivasan, Vigneshwari
• Format: Journal Article
• Language: English
• Published: Hoboken, USA John Wiley & Sons, Inc 15.02.2023; Wiley Subscription Services, Inc
• Subjects: Algorithms; Classification; Deep learning; deep maxout network; Disease control; Errors; Farms; Gas solubility; Machine learning; Mean square errors; Multilayers; Optimization; Optimization algorithms; Plant diseases; Recurrent neural networks; Resistance spot welding; Rice; rice plant; shuffled shepherd optimization algorithm; social optimization algorithm
• ISSN: 1532-0626; 1532-0634
• DOI: 10.1002/cpe.7523

Summary The earlier diagnosis and classification of plant diseases has the ability to control the spread of illnesses on a variety of crops with the aim of improving crop quality and yield. The automatic system effectively recognizes the plant diseases at less error and cost without the interpretation of farm specialists. In this article, shuffled shepherd social optimization‐based deep learning (SSSO‐based deep learning) technique is developed to classify rice leaf disease and severity percentage prediction. The classification is carried out using deep maxout network and the severity percentage prediction is performed using deep LSTM. The training of both deep learning techniques is achieved using developed SSSO algorithm, which is the combination of shuffled shepherd optimization algorithm (SSOA) and social optimization algorithm. The proposed technique achieved maximum accuracy, sensitivity, specificity of 0.926, 0.935, 0.892, and minimum mean square error, and root mean square error of 0.106, and 0.326. The accuracy of the implemented approach is 7.24%, 5.29%, 4%, and 2.81% improved than the existing techniques, like bacterial leaf streak‐based UNet (BLSNet), multilayer maxout, resistance spot welding‐based deep recurrent neural network (RSW‐based deep RNN), and rider Henry gas solubility optimization_deep neuro fuzzy network (RHGSO_DNFN) + deep LSTM.