Diagnóstico automático de Ceratoconjuntivite Infecciosa Bovina por meio de Imagens Termográficas e Redes Neurais Convolucionais

Abstract

Infectious bovine keratoconjunctivitis (IBK) is considered the most important ocular disease in cattle breeding, causing losses in the herd and damage to the producer. Because of being an infectious disease it is necessary to improve diagnostic forms. Currently, the diagnosis for IBK is made by evaluating clinical signs by a specialist and confirmed by laboratory tests, which is usually a costly task. In this work, the use of infrared thermography is evaluated for the acquisition of images of the bovine ocular region of healthy and experimentally infected animals by IBK. The available image base was previously classified by a specialist and subsequently used in the training and validation process of a set of distinct architectures of convolutional neural networks (CNN), which were evaluated using cross-validation. The best results indicate that CNN are able to identify cib clinical signs in thermographic images with accuracy close to 94%.