Desarrollo de un sistema portable de procesamiento de datos obtenidos por cámara para generar datasets de figura humana para los profesores de la Universidad Politécnica Salesiana

Abstract

The system is capable of meeting the need for datasets for training person recognition systems, whether through facial recognition or additional information obtained from the human figure. The collected information comes from an FPGA-type integrated circuit with a camera sensor, programmed with software developed in Python, capturing real-time images from the environment where it is placed. Subsequently, it processes these images to identify the figure or figures within the camera's focus, determining whether they are facing it or not, and finally separating their faces. All this information is stored in folders organized for each time a person enters the camera's focus. Then, it undergoes a process to identify which individuals were in front of the camera more than once, grouping their folders through a facial comparison. This results in a precise dataset, practically ready for training other systems in the field related to computer vision.