Media Memorability 2020 dataset
收藏arXiv2021-12-04 更新2024-06-21 收录
下载链接：
https://doi.org/10.6084/m9.figshare.15105867.v2
下载链接
链接失效反馈官方服务：
资源简介：
Media Memorability 2020数据集由埃塞克斯大学等机构创建，包含从TRECVid 2019 Video-to-Text数据集中选取的1500个短视频，分为训练、开发和测试集。数据集通过在线记忆游戏收集了用户对视频的短期和长期记忆标注，涉及1275名用户。每个视频附有文本描述和图像级特征，用于训练机器学习模型预测视频记忆性。该数据集应用于MediaEval 2020基准测试，旨在解决视频记忆性的自动计算问题，对广告和教育视频制作等领域具有重要价值。
The Media Memorability 2020 dataset was created by the University of Essex and other institutions. It contains 1500 short videos selected from the TRECVid 2019 Video-to-Text dataset, split into training, development, and test sets. Short-term and long-term memory annotations of the videos from users were collected via an online memory game, involving 1275 participants. Each video is accompanied by text descriptions and image-level features, which are used to train machine learning models for predicting video memorability. This dataset was applied to the MediaEval 2020 benchmark task, aiming to address the automatic computation of video memorability, and has important value in fields such as advertising and educational video production.
提供机构：
埃塞克斯大学, 英国
创建时间：
2021-12-04
