Tortelli, Daniel MeninKolassa, Caroline Paula2024-02-022024-02-022023https://repositorio.uricer.edu.br/handle/35974/479With the constant growth of the vehicle fleet worldwide, traffic safety has become an increasingly recurring concern. Among the causes of accidents in this field, the lack of attention of drivers to the signs on public roads stands out, which can lead to serious accidents. In this con text, this work proposes to train a neural network with the object detection algorithm YOLO in version 8 to identify 15 regulatory and 2 warning signs. The Roboflow framework was used for image labeling and the Python programming language for model training. The training and study carried out on the subject contributes to the growth of knowledge in an area that is so important for the development of technologies that can save lives on the road. The average result of the trained network reached 69.9%, the mAP reached 65.3%, and the Recall was 58.8%, fulfilling the proposed identification objective.pt-BRCiência da ComputaçãoRede neuralSegurança no trânsitoInteligência artificialTrabalho de Conclusão de Curso