GameScope: A Multi-Attribute, Multi-Codec Benchmark Dataset for Gaming Video Quality Assessment

2026-05-02

AI Summary

This work addresses the lack of large-scale subjective datasets covering multiple video codecs and content types in game video quality assessment, which has hindered the development of generalizable models. We present the largest game video quality dataset to date, encompassing 4,048 video samples encoded with H.264, H.265, and AV1, each annotated with an average of 37 subjective mean opinion scores (MOS) and multidimensional coarse-grained quality attributes. Through extensive subjective experiments, we enable, for the first time, cross-codec and cross-content-type evaluation of game video quality and introduce multidimensional quality attributes to support perceptual factor analysis. Experimental results demonstrate that our dataset significantly outperforms existing benchmarks, and that vision-language models trained on it surpass conventional approaches, establishing a robust foundation for future quality assessment research.

Abstract

The development of video game streaming has grown rapidly, with major platforms such as YouTube and Twitch using different codecs. To support quality assessment models that work consistently across any codec, it is necessary to have access to large, diverse subjective gaming quality datasets. Currently, there are only a few available, each having limitations. To address this gap, we present the largest gaming video quality dataset to date, incorporating both user-generated content (UGC) and professional-generated content (PGC) with extensive visual diversity. Our dataset covers the most widely used codecs - H.264, H.265, and AV1 - and consists of 4,048 video samples, each annotated by an average of 37 mean opinion score (MOS) ratings. In addition to overall quality scores, we collect coarse-grained quality attributes, enabling a better understanding of perceptual factors. We study the performance of leading video quality assessment methods on this dataset, including a vision language model that outperforms all the benchmarks. To the best of our knowledge, this is the first dataset that comprehensively addresses gaming video quality assessment across multiple codecs and content types with quality attributes. Our dataset is publicly available at https://rajeshsureddi.github.io/GameScope/.

Problem

• gaming video quality assessment
• multi-codec
• subjective dataset
• quality attributes
• video streaming

Innovation

• gaming video quality assessment
• multi-codec benchmark
• subjective quality dataset
• quality attributes
• vision language model