矩阵故障！基于内容驱动的音视频伪造检测与定位的大规模基准测试（英文版）

英文标题："Glitch in the Matrix!": A Large Scale Benchmark for Content Driven Audio-Visual Forgery Detection and Localization中文摘要：本文提出了一种用于检测 Deepfake 的方法，它通过提出包含多种模式的策略内容驱动音频、视觉和音视频混合进行识别。并通过定量分析证明了 BA-TFD + 算法再 Deepfake 检测方面的优越性。英文摘要：Most deepfake detection methods focus on detecting spatial and/orspatio-temporal changes in facial attributes. This is because availablebenchmark datasets contain mostly visual-only modifications. However, asophisticated deepfake may include small segments of audio or