Grape Leaf Disease Prediction Using Computer Vision and Deep Learning

Abstract

Human beings depend on plants for food. It also protects the earth from global warming and gives us rain. So, growing and saving plants are necessary and in recent days the yield is minimized. The main reason for the loss in agriculture is plant diseases. In the early days, a disease that affected the plant can only be identified by experts but it takes more time. The wrong prediction of the disease that affected the plant leads to incorrect use of pesticides on affected plants. At last, the farmers suffer from loss due to the incorrect usage of pesticides. So, an accurate prediction of the disease that affected the plant is necessary for avoiding other plants from being infected and also loss. Computer vision is used in this paper to identify the disease that affected the plant. In some cases, the disease identified by the computer can be wrong if the data given for the computer is not sufficient. In the proposed system deep learning is used to train the computer using the neural network models and let the computer make the decision so that it can predict the disease with accuracy and the accuracy of the model is 97.27%. The neural network model is built from scratch. The publicly available grapes leaf dataset is collected. The dataset contains the images of diseased leaves with labels and healthy leaves with labels to classify whether the leaf given as input is healthy or unhealthy. The following are steps followed to detect the leaf disease Image Acquisition, Image Preprocessing, Feature Extraction and Classification.