Classification of walnut diseases with deep learning

Walnut is an important plant grown for both food and wood industry in our country. Some diseases and mineral deficiencies may occur in walnut with the effect of various environmental factors. These diseases and mineral deficiencies manifest themselves in different ways in walnut leaves. Walnut producers are getting help from new generation technologies to combat some diseases and nutritional deficiencies that directly affect walnut productivity. In recent years, deep learning methods have been used to solve various problems in the agricultural sector as in many sectors. In this study, Faster R-CNN model was developed for the detection and classification of walnut anthracnose and vitamin deficiencies in walnut leaves. For the developed model, a dataset consisting of four classes, one of which is healthy and the other three are diseases, was used. The developed model was also tested with some pre-trained deep learning models such as GoogleNet, AlexNet and SqueezeNet and its classification success was analysed. For each CNN model, the effect of different iteration numbers on classification success was examined with Adam and SGDM optimisation algorithms. Metric values such as accuracy, precision, recall and F1 score of the models were calculated and their consistency was evaluated. The classification accuracy of the Faster R-CNN model developed in the study was found to be 98.28%.