Tomato Leaf Disease Detection using Neural Networks

Plants are a very important source of food worldwide. The diseases among them generate production and economic losses leading to a decrease in agricultural product quality and amount. Disease watching by hand is long and susceptible to mistakes. One of such plants is the Tomato plant, the foremost wide consumed vegetable in the Asian nation. Early illness detection is to scale back future losses. It is so crucial that a Convolutional Neural Network (CNN) with two convolutions and two max-pooling layers, together flattening layer followed by two totally connected layers is employed during this study in order to diagnose herbaceous plant diseases. Pictures of tomato leaves with 9 diseases and healthy samples were accustomed to ensure the findings. The images area unit metameric mistreatment image scaling, color threshold and flood filling techniques are used for findings. With our classification drawback mistreatment the notion of transfer learning. Within the tomato crop, nine of the foremost common leaf diseases and one healthy leaves were chosen for categorization. The projected model has a mean accuracy of 91.2 % for the 9 disorders and one healthy category.