Implementação de um aplicativo iOS baseado em processamento de imagens com frameworks nativas.

Abstract

Technological advances have provided more and more resources for the agricultural sector. The application of Artificial Intelligence (AI) to create solutions that help detect and diagnose animal diseases can be a valuable resource for reducing losses and minimizing production costs. AI, especially its sub-area of ML, enables the development of models capable of identifying patterns that can be used to automatically diagnose diseases. In the context of sheep and goat farming, methods are being used to improve the efficiency of parasite treatment in order to mitigate the increase in parasite resistance to anthelmintic drugs. The FAMACHA© method is widespread and effective in detecting anemia in sheep. However, it relies on human interpretation, which can result in errors. In order to automate the process of assessing the health of these sheep, this study describes the creation of a mobile application, implemented in Swift for iOS systems. The application uses exclusively Apple frameworks and libraries, which allow complex AI concepts to be abstracted. This makes it possible for developers who don’t work in this specific field to create intelligent applications. To offer a diagnosis of the animal based on the dominant color extracted from its ocular mucosa, 200 photographs of animals with their mucous membranes exposed were used. About two-thirds of these photographs were of non-anemic animals, while one-third were of anemic animals. The result was an application with satisfactory performance, capable of detecting the animal’s mucosa and classifying it.