Kawari - desenvolvimento de uma aplicação móvel aokucada a identificação de doenças foliares em culturas agrícolas

Abstract

The growing need to address agricultural losses caused by pests and diseases, combined with the increasing demand for food, drives the search for new techniques and technologies to assist farmers. Motivated by the need to help small farmers with the rapid and accurate identification of foliar diseases in plants, an application named KAWARI was developed using Computer Vision and Artificial Neural Networks to diagnose leaf diseases in plants. The app uses Computer Vision and Artificial Neural Networks to analyze images captured by the user and provide a diagnosis, along with additional information about the disease, nearby focus points, treatment products, and local suppliers. For model training, the PlantVillage dataset was used, containing images of diseased leaves from apple trees, grapevines, cherry trees, and strawberry plants. To enhance the model’s robustness, image preprocessing techniques such as brightness and contrast adjustments, rotation, and zoom were applied to augment the dataset. Additionally, a Node.js API was developed to handle communication between the application and the neural network, alongside a mobile application built with React Native. The communication between the application, API, and neural network proved efficient, with the system achieving a 98% accuracy rate in image diagnosis. This work demonstrates the potential of Computer Vision as a tool to support agriculture, enabling precise and rapid detection of plant diseases.