Aprendizaje Basado en Proyectos para la enseñanza de Tensores: Una estrategia didáctica aplicada a la Visión Artificial Profunda

Abstract

Abstract: This research developed, implemented, and validated a methodology based on Project-Based Learning (PBL) for teaching tensors applied to deep computer vision. Given the difficulties students had in understanding and applying advanced mathematical structures in programming environments, a pre-experimental design with pre- and post-tests in a single group was adopted, applying instruments validated by expert judgment and reliability analysis. The model was implemented at the Technical University of Cotopaxi and integrated mathematical theory, computer vision algorithms, and coding in Jupyter Notebook through sequential projects. The results showed significant improvements in theoretical understanding, computational application, and accuracy, which were confirmed using the Wilcoxon test. In addition, students had a positive perception of the methodology. It was concluded that PBL is effective in teaching abstract topics in engineering, as it helps to better connect theory with practice. The proposed methodology is replicable and transferable to other technical-computational courses, representing a valuable contribution to the teaching of mathematics and artificial intelligence.