Aplicación de reconocimiento facial para el control de asistencia de estudiantes mediante modelos de visión artificial

Abstract

Facial recognition systems allow facial characteristics, emotions, biometric data, among others, to be detected; This technology has been implemented in society, in different areas such as: cell phone unlocking, access control, allowing you to perform everyday tasks more quickly. Educational institutions also adopt these new technologies to improve the services they offer and optimize processes. For this reason, the purpose of this project was to develop a facial recognition web application that will be used by the teacher to record attendance, using computer devices, defining the cell phone as the main tool, because it allows a better capture resolution. To develop the prototype, the CRISP-DM methodology was used, which consists of the phases, understanding the business, data analysis, construction of a face database, training the model with AI techniques using an image dataset, choosing the best trained model and finally the deployment, in addition the XP methodology was used for the development of the web application with its phases, planning, design, coding and testing. Accuracy and precision metrics were applied based on a confusion matrix that was obtained when evaluating AI models for facial recognition, taking into account the parameters of light, position, and resolution of the image to be examined. This made it possible to select the Face-Api.js model. Being the one with the greatest precision and accuracy for the development of the project, the ISO 9126 quality standard was also applied with the purpose of evaluating the functionality, reliability, usability and efficiency of the web application. In the end, it was found that the application successfully fulfilled the main objective of identifying students through a group photograph with great precision, but the response times are similar to the current attendance record.