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

JunKyu Lee; Blesson Varghese; Hans Vandierendonck

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

JunKyu Lee^*, Blesson Varghese, Hans Vandierendonck

^*Corresponding author for this work

School of Computer Science

Research output: Contribution to conference › Paper › peer-review

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 language	English
Publication status	Published - 28 Oct 2022
Event	IEEE/CVF Winter Conference on Applications of Computer Vision - Waikoloa, United States Duration: 3 Jan 2023 → 7 Jan 2023 https://wacv2023.thecvf.com https://doi.org/10.1109/WACV56688.2023

Conference

Conference	IEEE/CVF Winter Conference on Applications of Computer Vision
Abbreviated title	WACV 2023
Country/Territory	United States
City	Waikoloa
Period	3/01/23 → 7/01/23
Internet address	https://wacv2023.thecvf.com https://doi.org/10.1109/WACV56688.2023

Access to Document

https://arxiv.org/abs/2210.16083

1 Conference contribution

ROMA: run-time object detection to maximize real-time accuracy
Lee, J., Varghese, B. & Vandierendonck, H., 6 Feb 2023, Proceedings 2023 IEEE/CVF winter conference on applications of computer vision (WACV): 3-7 January 2023, Waikoloa, Hawaii. Berg, T. & Farrell, R. (eds.). Piscataway, NJ: IEEE, p. 6394-6403 10 p. 10030225. (2023 IEEE/CVF winter conference on applications of computer vision (WACV)).
Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Open Access
File

Cite this

@conference{0d966ec8c7e0499cbcb3740e3a1ef0f1,

title = "ROMA: run-time object detection To maximize real-time accuracy",

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.",

author = "JunKyu Lee and Blesson Varghese and Hans Vandierendonck",

year = "2022",

month = oct,

day = "28",

language = "English",

note = "IEEE/CVF Winter Conference on Applications of Computer Vision, WACV 2023 ; Conference date: 03-01-2023 Through 07-01-2023",

url = "https://wacv2023.thecvf.com, https://doi.org/10.1109/WACV56688.2023",

}

TY - CONF

T1 - ROMA: run-time object detection To maximize real-time accuracy

AU - Lee, JunKyu

AU - Varghese, Blesson

AU - Vandierendonck, Hans

PY - 2022/10/28

Y1 - 2022/10/28

N2 - 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.

AB - 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.

M3 - Paper

T2 - IEEE/CVF Winter Conference on Applications of Computer Vision

Y2 - 3 January 2023 through 7 January 2023

ER -

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

Abstract

Conference

Access to Document

Fingerprint

Research output

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

Cite this