Research Article
An Intelligent Edge System for Face Mask Recognition Application
@INPROCEEDINGS{10.1007/978-3-031-08878-0_8, author={Tuan Le-Anh and Bao Nguyen-Van and Quan Le-Trung}, title={An Intelligent Edge System for Face Mask Recognition Application}, proceedings={Industrial Networks and Intelligent Systems. 8th EAI International Conference, INISCOM 2022, Virtual Event, April 21--22, 2022, Proceedings}, proceedings_a={INISCOM}, year={2022}, month={6}, keywords={Edge computing IoT AI Docker/Containerd Kubernetes DevOps CI/CD Cluster management Monitoring}, doi={10.1007/978-3-031-08878-0_8} }- Tuan Le-Anh
Bao Nguyen-Van
Quan Le-Trung
Year: 2022
An Intelligent Edge System for Face Mask Recognition Application
INISCOM
Springer
DOI: 10.1007/978-3-031-08878-0_8
Abstract
In the modern age, the growth of embedded devices, IoT (Internet of Things), 5G (Fifth Generation) and AI (Artificial Intelligence), has driven edge AI applications. Adopting Edge computing for AI applications intends to deal with power consumption, network capacity, response latency issues. In this paper, we introduce an intelligent edge system. It aims to assist with managing and developing microservices based AI applications on embedded computers with limited hardware resource. The proposed system uses Docker/Containerd and lightweight Kubernetes cluster (K3s) for high availability, self-healing, load balancing, scaling and automated deployment. It also facilitates GPU (Graphics Processing Unit) to speed up AI applications. The centralized cluster management and monitoring features simplify clusters and services administration, especially on a large scale. Meanwhile, container registry and DevOps platform with built-in code repository and CI/CD (Continuous Integration/Continuous Delivery) offer continuous integration and delivery for AI applications running on the cluster. This improves the process of AI applications development and management at the edge. In this experience, we implement the face mask recognition application with the proposed system. This application engages the state-of-the-art and lightweight object detection models with deep learning, observing mask violations to contribute to reducing the spread of COVID-19 disease.
