Sistema de detección, extracción y reconocimiento de rostros en escenas de máximo 4 personas, para aplicaciones de videovigilancia residencial utilizando herramientas de software libre, en lugares cerrados
Fecha
Autor corporativo
Título de la revista
ISSN de la revista
Título del volumen
Editor
Compartir
Director
Altmetric
Resumen
The protection of people and their property is a point that seeks to improve day after day, the advancement of technology allows the field of electronic security to be increasingly effective to provide comfort to users, removing that mental burden of be thinking that they or their property will be affected by unscrupulous people, the development of security systems has advanced a lot, and one of the newest fields and that has manifested this advance is facial recognition, said field, provides the ability to detect and recognize human faces on video automatically, generating an alarm or control signal, either to alert the person or allow entry. Current developments that have the feature of recognizing multiple faces on video tend to have high costs, because they are implemented on platforms that are not based on free software, which increases the cost due to pay-per-use licenses. This, there is a shortage of studies that demonstrate how efficient is a system that is developed in free software, and that has facial detection when interacting with multiple faces, that is reliable for the protection of assets, and that has the capacity to identify people who wish to access places of residence without having the respective authorization. In this research, a theoretical contextualization of the types of algorithms and libraries for facial detection and recognition that exist in Python will be found in the first instance, these libraries are called OpenCV and Face-Recognition, since there is no study that shows maximum detection distance, execution times and operating angles, among others, in this investigation it was decided to test both algorithms to determine which one is more effective or which one best fits the video surveillance systems proposed in this investigation. Next, the algorithm proposed by the researchers will be shown, which merges different programs and libraries that allow the detection, extraction and recognition of multiple faces in real time, in addition, the graphical user interface (GUI) that allows total manipulation will be displayed. of this system, finally, the tests carried out on the system in two different environments, one residential and one commercial, will be shown. It should be clarified that the entire system was developed in the Python programming language, which is free.