UnFaZeD07/AVE-Dataset

AVE数据集是从GitHub项目移植而来的原始数据集，用于音频-视觉事件定位研究。数据集包含多个视频样本，每个视频可能包含不同的音频-视觉事件。注释文件（annotations.txt）提供了每个样本的事件类别、YouTube ID、质量（均为良好，表示包含音频-视觉事件）、音频-视觉事件的开始和结束时间。此外，数据集还提供了训练集、验证集和测试集的划分文件（train/val/test-Set.txt），这些划分在原始论文中使用。

The AVE dataset is the original dataset ported from a GitHub project, designed for audio-visual event localization research. The dataset includes multiple video samples, each potentially containing different audio-visual events. The annotations file (annotations.txt) provides the event category, YouTube ID, quality (all good, indicating the presence of an audio-visual event), and the start and end times of the audio-visual event for each sample. Additionally, the dataset includes split files for the training, validation, and test sets (train/val/test-Set.txt), which were used in the original paper.