2021-02-252021-02-252021-02-252020-12-11https://ri.ucsal.br/handle/prefix/3769Cataract is one of the diseases of the eyeball that generates the most blindness in the world. The rapid detection of this disease, close to the appropriate treatment, contributes to the improvement of the patients’ quality of life. This work uses convolutional neural networks to train three sets of data containing images of different patients, in order to classify cataract patients and non-carriers. For this, a convolutional neural network vgg19 had its architecture modified and improved. The most successful results used by the network, was an accuracy of 100 %.Acesso AbertoAprendizagem profundaRedes neurais convolucionaisCatarataDeep learningConvolutional neural networkCataractTrabalho de Conclusão de CursoEngenhariasEngenharia de Software