Cloud-Based Face Recognition for Augmented Reality Glasses
Journal
Proceedings - 2023 IEEE International Symposium on Mixed and Augmented Reality Adjunct, ISMAR-Adjunct 2023
ISBN
9798350328912
Date Issued
2023-01-01
Author(s)
Bedjo Rahardjo, David Akas
Abstract
Traditional face recognition systems are built to be deployed within a geographically constrained environment using a combination of cameras, processing units, and display monitors. However, this can limit the mobility and operational area of these systems. A wireless system with wearable display offers a potential solution to this problem. To address this issue, we propose a cloud-based face recognition system with augmented reality (AR) glasses. The system consists of three main components: AR glasses to capture images of a scene and display recognition results, mobile phone to operate the AR glasses and communicate with the server via Internet connection, and server to run a deep learning model to perform face recognition on images received. Our system allows users to connect to our server through the Internet, such as cellular networks, as well as public and local WiFi networks. Additionally, experimental results indicate that our system can simultaneously recognize up to two individuals while maintaining an average frame rate of approximately 14 frames per second, sufficient for enabling users to move around and constantly identify individuals in a scene.
Subjects
augmented reality | cloud computing | deep learning | Face recognition | Internet of Things
Type
conference paper