The 5G Vision project - Applications for Intelligent Campus over 5G private networks, conceived by RNP in partnership with Inatel and researchers from the Federal University of Rio de Janeiro (UFRJ), was approved in the public selection “Development of Applications in Private 5G Networks”, carried out through the Financier of Studies and Projects (Finep).  

The aim of the project is to develop a computer vision system that uses images from 5G cameras, associated with the use of IoT (Internet of Things) devices, for use in various applications on university campuses, such as detection and recognition of people for access control, automatic list of presence in classrooms, recognition of behavior, among others. The low-cost 5G system to be developed in the project will offer access to high communication rates, low latency, as well as the connection of various devices.   

“In recent years, we have observed a substantial growth in the use of artificial intelligence for detection and identification of people with the purpose of using it in real-time applications for monitoring and security”, explains José Ferreira Rezende, professor and researcher at COPPE/UFRJ and RNP R&DI advisor. “However, these applications still do not reach the real-time indicator for decision-making, as they have high latency to transfer the captured images, in addition to the need for a high number of frames to achieve good accuracy”, he adds.  

In this sense, the main objective of the application will be to have efficient data compression/decompression algorithms at the edge, in addition to performing facial and behavior recognition of the analyzed individuals.  

The facial recognition solution associated with access control will also have immediate application in places that require security, allowing only authorized people recognized by the system to have access. The function of recognizing the behavior of people, more specifically, of students in the classroom, aims to identify the level of attention of students during the course of the class, as well as the teaching techniques that most attract attention in these spaces.  

“These mechanisms will allow professors, course coordinators and pedagogues to evaluate the different teaching techniques, the performance of professors and students within the classroom, and others”, says Rezende. “The behavior recognition solution, if used by educational institutions, can generate evaluation and performance metrics for the continuous improvement of classes, generating real-time feedback from the target audience”.  

The researcher also believes that the advances achieved in the development of the computer vision system can be applied not only in classrooms, but in any situation or event that is composed of a presenter and an audience.  

The expected period for the execution of the 5G Vision project, which will cover the Southeast and Midwest regions, is 24 months.