Multiple Leaf Disease Detection by Hybrid Machine and Deep Learning SVM-CNN Model

Abstract

Identity of the plant illnesses is the important thing toward stopping the losses within the yield and amount of the agricultural product. Because the plant life be afflicted by the disease, the production of crop decreases because of infections due to several sorts of illnesses on its leaf, crop, and branch. Leaf sicknesses are mainly as a result of bacteria, fungi, virus and so forth. Illnesses are frequently tough to manipulate. Analysis of the sickness should be done appropriately and proper movements must be taken at the correct time. A correct detection of leaf disorder is crucial for plant culture as well as the rural financial system. Even though many works were executed for identifying leaf disease, due to the inadequate strategies additionally the obligations about classifying leaf disorder are difficult to be expecting. Here, in our proposed paper, pre-processing is carried out to clear the image using Adaptive Adjustment Algorithm (AAA), and then threshold-based segmentation Otsu`s method is proposed to segment the based-on threshold values and then Resnet-50 based Convolutional Neural Network (CNN)transfer learning is proposed to extract the features and then Support Vector Machine (SVM) classification technique is used to classify the various leaf disease grade with plant crops. Finally, the performance of our proposed model gives the better accuracy 99.98% using our proposed model compared with other techniques.