Identificação de cédulas falsas utilizando aprendizado profundo

Abstract

The spread of counterfeit currency has become a pressing concern for authorities worldwide, as such practices have a detrimental impact on the economy and result in significant financial losses. Furthermore, the advancement of modern technologies has made the production of forgeries more accessible. In the face of this scenario, the primary objective of this research is to develop a deep learning model for the classification of counterfeit banknotes, using a set of images representing these notes as input. The adopted methodological approach involves the use of a database comprising images of Colombian peso banknotes, followed by a preprocessing step. Subsequently, classification models were developed employing convolution neural networks. After implementation, the models were evaluated. The results demonstrate that the implemented models achieved highly satisfactory performance in the task of distinguishing between counterfeit and authentic notes, with final accuracy of 100%.