Classification of Foliar Diseases in Apple Tree using Neural Network

The development of highly resistant pathogenic strains, high production costs, and misdiagnosis of various crop diseases can result in the abuse of pesticides, which significantly negatively affects the economy and the environment. Making an efficient disease control strategy helps to avoid financial losses. Computer vision has been able to detect the fungal infection on leaf by the computer vision frameworks that can enhance the performance and their techniques are developed through deep learning. Human scouting is currently an expensive and time-consuming method of illness diagnosis. In this study, a structure of a convolutional neural network has been proposed that is capable of diagnosing disease in apple plant leaves. We compared the DenseNet CNN, CNN + XGBOOST, CNN + Random Forest, and CNN + SVM and according to the results, neural networks are still able to give better accuracy than the rest of the proposed models with an accuracy of 99%.