Plant Leaves Diseases Recognition using Convolutional Neural Network and five other Hybrid Models

Due to the unavailability of up-to-the-mark advancements in the agriculture sector, crop production suffers a lot. In order to prevent crops from various diseases, early and correct diagnosis of diseases occurred have its supreme significance. In this paper, we have considered various infectious plant diseases with reference to different plants available. All the data are taken from the publicly available dataset. Along with CNN (Convolutional Neural Network) algorithm with the accuracy of 94.83%, we have implemented five other hybrid supervised machine learning classification algorithms for classification of various diseases, namely, CNN+SVM (CNN + Support Vector Machines) with the accuracy of 96.87%, CNN+DT (CNN + Decision Trees) with the accuracy of 15.97%, CNN+KNN (CNN + K-Nearest Neighbor) with the accuracy of 96.54%, CNN+RFC (CNN + Random Forest Classifier) with the accuracy of 99.93% and CNN+NB (CNN + Naïve Bayes) with accuracy of 94.8%. With the accuracy of 99.93%, CNN+RFC hybrid model gives the best results so far.