Tomato Leaf Disease Detection System Using Convolutional Neural Networks
Samkeliso Suku Dube, Elizabeth Chiwera, Sindiso Nleya, Siqabukile Ndlovu, Tsitsi Zengeya, Namatirai Marabada, Joseph Mutengeni, Wellington Mapenduka
Abstract
The tomato leaf disease detection system using convolutional neural networks (CNN) is a computer visionbased approach to automate the identification of tomato leaf disease. Convolutional neural network is an artificial neural network, which is used widely in prediction. The system is a web-based application that allows farmers to upload images then get real-time results on whether the leaf uploaded is healthy or not. The system also prescribes detected disease treatment options. In Zimbabwe, leaf diseases in tomatoes are increasingly becoming common. The diseases affect the quality and quantity of tomato fruit output leading to poor yields and low monetary return on Agri-investments. Tomato leaf diseases are common problems in the farming industry, mostly in Zimbabwe as the field has become one of most paying fields, these problems affect the quality and quantity of tomato plants. The research aimed to develop a classifier that in real time predicts disease results from a tomato leaf image upload. The dataset of tomato leaves was downloaded from Kaggle, the dataset was used to train the classification model as well as testing the model for prediction accuracy. The study involved the collection of the dataset of tomato leaf images, pre-processing the images, training the CNN model and developing the web application. The accuracy of the system's performance was 93% which is a good result in that the farmers can rely on the system's classification.