QoS-oriented Service Management in clouds for large scale industrial activity recognition

Motivated by the need of industrial enterprises for supervision services for quality, security and safety guarantee, we have developed an Activity Recognition Framework based on computer vision and machine learning tools, attaining good recognition rates. However, the deployment of multiple cameras...

Full description

Saved in:
Bibliographic Details
Main Authors: Voulodimos, A. S., Kyriazis, D. P., Gogouvitis, S. V., Doulamis, A. D., Kosmopoulos, D. I., Varvarigou, T. A.
Format: Conference Proceeding
Language:English
Subjects:
activity recognition
Cameras
cloud infrastructure
Computer vision
Hidden Markov models
industrial workflows
Monitoring
Pattern recognition
QoS
Quality of service
Real time systems
service management
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Motivated by the need of industrial enterprises for supervision services for quality, security and safety guarantee, we have developed an Activity Recognition Framework based on computer vision and machine learning tools, attaining good recognition rates. However, the deployment of multiple cameras to exploit redundancies, the large training set requirements of our time series classification models, as well as general resource limitations together with the emphasis on real-time performance, pose significant challenges and lead us to consider a decentralized approach. We thus adapt our application to a new and innovative real-time enabled framework for service-based infrastructures, which has developed QoS-oriented Service Management mechanisms in order to allow cloud environments to facilitate real-time and interactivity. Deploying the Activity Recognition Framework in a cloud infrastructure can therefore enable it for large scale industrial environments.
DOI:10.1109/SoCPaR.2011.6089156