ROMA: run-time object detection to maximize real-time accuracy

JunKyu Lee, Blesson Varghese, Hans Vandierendonck

Research output: Chapter in Book/Report/Conference proceedingConference contribution

8 Downloads (Pure)

Abstract

This paper analyzes the effects of dynamically varying video contents and detection latency on the real-time detection accuracy of a detector and proposes a new run-time accuracy variation model, ROMA, based on the findings from the analysis. ROMA is designed to select an optimal detector out of a set of detectors in real time without label information to maximize real-time object detection accuracy. ROMA utilizing four YOLOv4 detectors on an NVIDIA Jetson Nano shows real-time accuracy improvements by 4 to 37% for a scenario of dynamically varying video con-tents and detection latency consisting of MOT17Det and MOT20Det datasets, compared to individual YOLOv4 detectors and two state-of-the-art runtime techniques.
Original languageEnglish
Title of host publicationProceedings 2023 IEEE/CVF winter conference on applications of computer vision (WACV)
Subtitle of host publication3-7 January 2023, Waikoloa, Hawaii
EditorsTamara Berg, Ryan Farrell
Place of PublicationPiscataway, NJ
PublisherIEEE
Pages6394-6403
Number of pages10
ISBN (Electronic)9781665493468
ISBN (Print)9781665493475
DOIs
Publication statusPublished - 6 Feb 2023
EventIEEE/CVF Winter Conference on Applications of Computer Vision - Waikoloa, United States
Duration: 3 Jan 20237 Jan 2023
https://wacv2023.thecvf.com
https://doi.org/10.1109/WACV56688.2023

Publication series

Name2023 IEEE/CVF winter conference on applications of computer vision (WACV)
ISSN (Print)2472-6737
ISSN (Electronic)2642-9381

Conference

ConferenceIEEE/CVF Winter Conference on Applications of Computer Vision
Abbreviated titleWACV 2023
Country/TerritoryUnited States
CityWaikoloa
Period3/01/237/01/23
Internet address

Keywords

  • Applications: embedded sensing/real-time techniques

Fingerprint

Dive into the research topics of 'ROMA: run-time object detection to maximize real-time accuracy'. Together they form a unique fingerprint.

Cite this