Modelo de aforo automatizado para estudios de movilidad utilizando reconocimiento facial en el sistema integrado de transporte de Bogotá (sitp)

Abstract

This work carries out and develops a methodology for the generation of origindestination matrices in mobility studies, using advanced facial recognition tools. For this, a bibliographic compilation is carried out that covers two topics. Firstly, the origin-destination matrix generation models are analyzed, classifying their advantages and disadvantages in terms of precision, preparation time for mobility studies. Secondly, pre-trained facial recognition models are studied, evaluating their precision and capabilities in different contexts. The experimentation proposed in this work focuses on the application of facial recognition models for user registration during the loading and unloading of users in the Integrated Public Transport System (SITP). From this analysis, the most appropriate model is selected for this case, with DeepFace being the starting point for the implementation of codes that allow capturing and comparing faces. As a result, a facial recognition-based information capture methodology is proposed, aimed at generating origin-destination matrices in real time. This approach seeks to provide accurate origin-destination pairs, improving the accuracy and efficiency of urban mobility studies and opening up new possibilities for public transportation optimization.