Convolutional Neural Network with Improved Optimizer for Plant Disease Detection and Classification

Recently, plant disease detection has become a main concern in the field of agriculture. The early detection and identification of plant disease supports the farmers for considering precaution metrics by protecting the diseases spread to various phases of plants. The existing techniques to detect plant diseases majorly concentrated on manual feature extraction, making made it expensive. Here, the Convolutional Neural Network (CNN) with improved optimizer is developed to detect and classify the plant leaf disease with high classification accuracy. The deep meaningful features are automatically extracted by using the convolutional layer of CNN. Then, the improved Stochastic Gradient Descent (SGD) optimizer is used in the CNN model which enhances the classification performance. The performance measures considered to evaluate the performance of the developed method are, accuracy, f1-score, recall and precision. The CNN with improved optimizer attains 99.5% accuracy, 98.7% precision, 98.2% recall and 98.5% f1-score, proving superior to the existing techniques.