MovieCuts

Movicuts是视频和剪辑标签的大型数据集，包含173,967视频剪辑，这些视频剪辑被标记为10种不同剪辑类型的q。数据集中的每个样本由两个视频剪辑 (从中分割一个视频剪辑) 及其随附的音频组成。该数据集专注于一个新任务: 视频剪辑类型识别。视频编辑是指将两个视频拼接在一起，拼接后的视频看起来是连续的，不太引人注目。视频剪辑类型识别是识别两个视频拼接在一起的方式。Movicuts is a large-scale dataset focused on video clips and their editing tag annotations. It contains 173,967 video clips annotated with 10 distinct cut types. Each sample in the dataset consists of two video clips (derived from splitting a single source video) along with their accompanying audio tracks. This dataset targets a novel task: video cut type recognition. Video splicing refers to the technique of combining two video clips into a composite video that appears seamless and inconspicuous. Video cut type recognition is the task of identifying the specific manner in which two video clips are spliced together.

搜集汇总

背景与挑战

背景概述

MovieCuts是一个大型视频剪辑标签数据集，包含173,967个视频剪辑，标记为10种不同的剪辑类型，专注于视频剪辑类型识别任务。该数据集由Adobe Research和King Abdullah University of Science and Technology于2022年发布。