No abstract

Tomato is an economically important crop plant in India. However, pests, diseases and post-harvest loss makes its cultivation very costly. To overcome one such issue, early detection of pests and diseases is the need of the hour which aids small-scale farmers, who make up a significant portion of the labor force in developing countries, in reducing economic losses. This study intends to assist farmers in making pest management decisions, reducing the usage of unnecessary pesticides and resulting in higher-quality goods. The fact that these illnesses and pests are primarily confined makes it difficult to classify distinct insect species on a large scale due to their high degree of similarity in traits. The accuracy of four architectures: CNN (Convolutional Neural Networks), SVM (Support Vector Machine) and CNN Architectures like ResNet152V2 and EfficientNetV2 is compared in this article. The models were trained using a library of tomato photos. After testing and training architectures, the model that performed best overall can be used in future development, such as developing an app that allows farmers to quickly detect disease. The suggested IoT can be initially applied in greenhouses which are much more confined in terms of space and hence can be effectively integrated.