ExDDV
收藏资源简介:
ExDDV是一个 Explainable Deepfake Detection in Video 的数据集,由罗马尼亚布加勒斯特大学和西罗马尼亚蒂米什瓦拉大学共同创建。该数据集包含约5.4K个真实和深伪视频,这些视频经过人工标注,带有文本描述和点击标记,以解释和定位视频中的异常。数据集从多个现有数据集中收集视频,以增强多样性。ExDDV官方分为训练集、验证集和测试集,方便结果复现和未来比较,可应用于视频深伪检测的可解释性研究。
ExDDV is a dataset for Explainable Deepfake Detection in Video, co-developed by the University of Bucharest in Romania and West University of Timișoara. This dataset contains approximately 5.4K real and deepfake videos, which are manually annotated with textual descriptions and clickable markers to explain and localize anomalies present in the videos. It aggregates videos from multiple existing datasets to enhance its diversity. Officially, ExDDV is partitioned into training, validation and test subsets to facilitate result reproducibility and cross-study comparisons, and can be applied to explainable research for video deepfake detection.




