Edge Computing Resource Management for Cross-Camera Video Analytics: Workload and Model Adaptation
Multi-camera systems are now widely employed across numerous domains. The exponential growth of deep learning has simplified the implementation of advanced video analytics applications. While current systems strive to enhance live video analytics from several aspects, they overlook the potential deg...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2024-01-01
|
Series: | IEEE Access |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/10403890/ |
_version_ | 1797335545023561728 |
---|---|
author | Huan-Ting Chen Yao Chiang Hung-Yu Wei |
author_facet | Huan-Ting Chen Yao Chiang Hung-Yu Wei |
author_sort | Huan-Ting Chen |
collection | DOAJ |
description | Multi-camera systems are now widely employed across numerous domains. The exponential growth of deep learning has simplified the implementation of advanced video analytics applications. While current systems strive to enhance live video analytics from several aspects, they overlook the potential degradation in performance resulting from dynamic content changes, such as the variation in the quantity and types of objects of interest over a period and across different cameras. The authors introduce Workload and Model Adaptation (WMA), a two-stage resource allocation strategy for a three-tiered, cross-camera video analytics system. This system not only supports model fine-tuning but also ensures workload balance. Notably, both the system architecture and control workflow fully comply with the IEEE 1935 edge standard. This paper delves into the GPU utilization performance of a vehicle re-identification application and examines the workload dynamics spanning multiple cameras. Furthermore, the challenges related to multi-process execution are explored. The system is evaluated using a commonly employed dataset and a popular open-source project. The results demonstrate that the proposed design surpasses the baseline and enhances the overall throughput and latency across cameras within the system. |
first_indexed | 2024-03-08T08:39:06Z |
format | Article |
id | doaj.art-3be874747f824f9dba7cc8e956dc3929 |
institution | Directory Open Access Journal |
issn | 2169-3536 |
language | English |
last_indexed | 2024-03-08T08:39:06Z |
publishDate | 2024-01-01 |
publisher | IEEE |
record_format | Article |
series | IEEE Access |
spelling | doaj.art-3be874747f824f9dba7cc8e956dc39292024-02-02T00:01:01ZengIEEEIEEE Access2169-35362024-01-0112120981210910.1109/ACCESS.2024.335581510403890Edge Computing Resource Management for Cross-Camera Video Analytics: Workload and Model AdaptationHuan-Ting Chen0https://orcid.org/0009-0007-2989-6129Yao Chiang1https://orcid.org/0000-0002-0392-6525Hung-Yu Wei2https://orcid.org/0000-0002-3116-306XDepartment of Electrical Engineering, National Taiwan University, Taipei, TaiwanDepartment of Electrical Engineering, National Taiwan University, Taipei, TaiwanDepartment of Electrical Engineering, National Taiwan University, Taipei, TaiwanMulti-camera systems are now widely employed across numerous domains. The exponential growth of deep learning has simplified the implementation of advanced video analytics applications. While current systems strive to enhance live video analytics from several aspects, they overlook the potential degradation in performance resulting from dynamic content changes, such as the variation in the quantity and types of objects of interest over a period and across different cameras. The authors introduce Workload and Model Adaptation (WMA), a two-stage resource allocation strategy for a three-tiered, cross-camera video analytics system. This system not only supports model fine-tuning but also ensures workload balance. Notably, both the system architecture and control workflow fully comply with the IEEE 1935 edge standard. This paper delves into the GPU utilization performance of a vehicle re-identification application and examines the workload dynamics spanning multiple cameras. Furthermore, the challenges related to multi-process execution are explored. The system is evaluated using a commonly employed dataset and a popular open-source project. The results demonstrate that the proposed design surpasses the baseline and enhances the overall throughput and latency across cameras within the system.https://ieeexplore.ieee.org/document/10403890/Continuous learningedge computingre-identificationresource allocationoffloadingvideo analytic |
spellingShingle | Huan-Ting Chen Yao Chiang Hung-Yu Wei Edge Computing Resource Management for Cross-Camera Video Analytics: Workload and Model Adaptation IEEE Access Continuous learning edge computing re-identification resource allocation offloading video analytic |
title | Edge Computing Resource Management for Cross-Camera Video Analytics: Workload and Model Adaptation |
title_full | Edge Computing Resource Management for Cross-Camera Video Analytics: Workload and Model Adaptation |
title_fullStr | Edge Computing Resource Management for Cross-Camera Video Analytics: Workload and Model Adaptation |
title_full_unstemmed | Edge Computing Resource Management for Cross-Camera Video Analytics: Workload and Model Adaptation |
title_short | Edge Computing Resource Management for Cross-Camera Video Analytics: Workload and Model Adaptation |
title_sort | edge computing resource management for cross camera video analytics workload and model adaptation |
topic | Continuous learning edge computing re-identification resource allocation offloading video analytic |
url | https://ieeexplore.ieee.org/document/10403890/ |
work_keys_str_mv | AT huantingchen edgecomputingresourcemanagementforcrosscameravideoanalyticsworkloadandmodeladaptation AT yaochiang edgecomputingresourcemanagementforcrosscameravideoanalyticsworkloadandmodeladaptation AT hungyuwei edgecomputingresourcemanagementforcrosscameravideoanalyticsworkloadandmodeladaptation |