GameVibe
收藏资源简介:
GameVibe是由马耳他大学的数字游戏研究所创建的一个新颖的多模态情感游戏语料库,专注于第一人称射击(FPS)游戏视频的观众参与度。该数据集包含从YouTube上的公开“Let’s Play”视频中提取和精选的30个不同FPS游戏的高质量音频和视觉数据,总计2小时。数据收集涉及20名标注者,包括训练有素的研究人员、研究生和本科生。情感标签以无界、时间连续信号的形式提供,使用RankTrace标注工具在PAGAN平台上进行。数据集包括原始视频、预训练基础模型提取的潜在表示以及每个标注者的质量保证数据,旨在提高下游任务中情感模型的一般化能力。
GameVibe is a novel multimodal affective gaming corpus created by the Digital Games Institute of the University of Malta, focusing on viewer engagement in first-person shooter (FPS) gameplay videos. This dataset contains high-quality audio and visual data from 30 distinct FPS games, extracted and curated from publicly available "Let's Play" videos on YouTube, with a total duration of 2 hours. The data collection involved 20 annotators, including trained researchers, graduate students, and undergraduate students. The affective labels are provided as unbounded, temporally continuous signals, annotated using the RankTrace annotation tool on the PAGAN platform. The corpus includes raw videos, latent representations extracted by pre-trained foundation models, and quality assurance data for each annotator, aiming to enhance the generalizability of affective models in downstream tasks.




