Artificial Intelligence in Education 5.0: a methodology for three-dimensional geometric shape classification for an educational tool

Traditional teaching methodologies have reflected in low educational indices in subjects such as mathematics in Brazil. It is in this context that Education 5.0 appears as a way to improve teaching allied to the use of technology. The purpose of this work is to integrate a model of geometric shapes classification to an educational application for the basic teaching of plane and spatial geometry involving the Metaverse. Therefore, this work proposes: (i) the fine-tuning of three-dimensional geometric shapes classification models; (ii) the evaluation of performance, inference time and model dimension; and (iii) the selection of the best performing model according to evaluation criteria based on the precision of each model and its inference performance. According to the evaluation criteria used and the results obtained, it was verified that the ResNet model presented the best results for integration with the educational application.