Q3Vision: A Vision-Based Framework for Game-State Reconstruction in Quake III Arena

Pham, DM; Kim, J; Franklin, D; Abolhasan, M; Owen, R

Q3Vision: A Vision-Based Framework for Game-State Reconstruction in Quake III Arena

Pham, DM Kim, J Franklin, D

Abolhasan, M

Owen, R

Permalink

Publisher:: Association for Computing Machinery (ACM)
Publication Type:: Conference Proceeding
Citation:: Proceedings of 2026 Australasian Computer Science Week Acsw 2026, 2026, pp. 100-108
Issue Date:: 2026-07-03

Open Access

Copyright Clearance Process

Recently Added
In Progress
Open Access

This item is open access.

Adobe PDF

Download Published versionAdobe PDF (4.73 MB)

View on publisher's site

View statistics

Full metadata record

Field	Value	Language
dc.contributor.author	Pham, DM
dc.contributor.author	Kim, J
dc.contributor.author	Franklin, D https://orcid.org/0000-0002-9563-5943
dc.contributor.author	Abolhasan, M https://orcid.org/0000-0002-4282-6666
dc.contributor.author	Owen, R https://orcid.org/0009-0002-0780-854X
dc.date.accessioned	2026-10-09T07:51:53Z
dc.date.available	2026-10-09T07:51:53Z
dc.date.issued	2026-07-03
dc.identifier.citation	Proceedings of 2026 Australasian Computer Science Week Acsw 2026, 2026, pp. 100-108
dc.identifier.uri	http://hdl.handle.net/10453/197175
dc.description.abstract	Computer vision offers new opportunities for understanding gameplay and player behavior in esports and interactive digital media. However, the performance of modern vision models in legacy first-person shooter (FPS) environments, characterized by rapid camera motion, motion blur, occlusion, and stylized heads-up display (HUD) elements, remains insufficiently studied. This paper presents Q3Vision, a vision-based framework for reconstructing game-state information in Quake III Arena using only screen-captured video. The system used YOLO-based object detection to identify enemies and interactable items, together with OCR-based extraction of player-status indicators such as health, armour, ammunition, and weapon state. A reproducible OBS-based data acquisition pipeline enables frame-level alignment of visual data and supports evaluation across varied gameplay conditions. A reproducible data acquisition pipeline based on Open Broadcaster Software (OBS) enables frame-level alignment of visual data and supports evaluation across varied gameplay conditions. Experimental results demonstrate that vision-only state reconstruction is feasible, though OCR performance is limited by low-resolution, stylized HUD fonts and visual noise. The findings highlight both the potential and the practical constraints of applying computer vision techniques to fast-paced FPS gameplay, informing future work on domain-specific OCR fine-tuning, temporal modeling, and multimodal extensions.
dc.language	en
dc.publisher	Association for Computing Machinery (ACM)
dc.relation.ispartof	Proceedings of 2026 Australasian Computer Science Week Acsw 2026
dc.relation.ispartof	Proceedings of the 2026 Australasian Computer Science Week
dc.relation.isbasedon	10.1145/3793811.3793821
dc.rights	info:eu-repo/semantics/openAccess
dc.title	Q3Vision: A Vision-Based Framework for Game-State Reconstruction in Quake III Arena
dc.type	Conference Proceeding
pubs.organisational-group	University of Technology Sydney
pubs.organisational-group	University of Technology Sydney/Faculty of Engineering and Information Technology
pubs.organisational-group	University of Technology Sydney/Faculty of Engineering and Information Technology/School of Electrical, Mechanical and Biomedical Engineering
pubs.organisational-group	University of Technology Sydney/UTS Groups
pubs.organisational-group	University of Technology Sydney/UTS Groups/Global Big Data Technologies Centre (GBDTC)
pubs.organisational-group	University of Technology Sydney/UTS Groups/Transport Research Centre (TRC)
pubs.organisational-group	University of Technology Sydney/UTS Groups/Cyber Digital Centre (CDC)
utslib.copyright.status	open_access	*
dc.rights.license	This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0). To view a copy of this license, visit https://creativecommons.org/licenses/by/4.0/
dc.date.updated	2026-10-09T07:51:52Z
pubs.publication-status	Published

Abstract:

Computer vision offers new opportunities for understanding gameplay and player behavior in esports and interactive digital media. However, the performance of modern vision models in legacy first-person shooter (FPS) environments, characterized by rapid camera motion, motion blur, occlusion, and stylized heads-up display (HUD) elements, remains insufficiently studied. This paper presents Q3Vision, a vision-based framework for reconstructing game-state information in Quake III Arena using only screen-captured video. The system used YOLO-based object detection to identify enemies and interactable items, together with OCR-based extraction of player-status indicators such as health, armour, ammunition, and weapon state. A reproducible OBS-based data acquisition pipeline enables frame-level alignment of visual data and supports evaluation across varied gameplay conditions. A reproducible data acquisition pipeline based on Open Broadcaster Software (OBS) enables frame-level alignment of visual data and supports evaluation across varied gameplay conditions. Experimental results demonstrate that vision-only state reconstruction is feasible, though OCR performance is limited by low-resolution, stylized HUD fonts and visual noise. The findings highlight both the potential and the practical constraints of applying computer vision techniques to fast-paced FPS gameplay, informing future work on domain-specific OCR fine-tuning, temporal modeling, and multimodal extensions.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/197175