Multilingual Synopses of Movie Narratives (M-SYMON)
收藏资源简介:
M-SYMON是由南洋理工大学计算与数据科学学院创建的大型多语言视频故事数据集,包含来自7种语言的13,166个电影摘要视频,总计2,136小时。数据集通过YouTube上的电影回顾视频收集,并手动注释了480个视频的精细视频-文本对应关系,总计101.5小时。M-SYMON旨在解决计算故事理解中的视频-文本对齐问题,特别是在多语言环境下的挑战,支持如文本到视频生成和视觉故事生成等应用。
M-SYMON is a large-scale multilingual video story dataset developed by the School of Computing and Data Science, Nanyang Technological University. It contains 13,166 movie recap videos across 7 languages, with a total duration of 2,136 hours. The dataset is collected from movie review videos on YouTube, and 480 of these videos have been manually annotated with fine-grained video-text correspondences, totaling 101.5 hours. M-SYMON aims to address the video-text alignment problem in computational story understanding, particularly the challenges in multilingual environments, and supports applications such as text-to-video generation and visual story generation.

- 1Multilingual Synopses of Movie Narratives: A Dataset for Story Understanding南洋理工大学计算与数据科学学院 · 2024年



