Image-Based Early Classification of Tomato Leaf Diseases Using Deep Neural Networks

Food for the world's population is largely produced by plants, however, plant diseases cause substantial output losses. Continuous monitoring can reduce these losses. However, manual disease monitoring of plants is labor-intensive and error-prone. Early disease diagnosis utilizing computer vision and artificial intelligence (AI) has emerged as a viable answer to this problem. This strategy lessens the detrimental effects of illnesses and gets around the drawbacks of ongoing human monitoring. After comparing the Convolutional neural network (CNN) with architectures such as ResNet, MobileNet, DenseNet, and InceptionV3, Cnn was applied to simple images of tomato leaves and classified into several tomato leaf diseases. We assessed the models' performance in a variety of ways, including binary classification (distinguishing between healthy and unhealthy leaves), and seven-class classification (categorizing healthy leaves and various groups of diseased leaves). After opting for Convolutional Neural Network (CNN), we were able to achieve a classification accuracy of 95.8%.